- developers - lists.mariadb.org

[Maria-developers] bzr commit into Mariadb 5.2, with Maria 2.0:maria/5.2 branch (igor:2745)
by Igor Babaev 19 Mar '10

19 Mar '10

#At lp:maria/5.2 based on revid:igor@askmonty.org-20100317023231-w7h0euroof0lul8e 2745 Igor Babaev 2010-03-18 Made the vcol suite independent on time zone. modified: mysql-test/suite/vcol/inc/vcol_supported_sql_funcs_main.inc mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result === modified file 'mysql-test/suite/vcol/inc/vcol_supported_sql_funcs_main.inc' --- a/mysql-test/suite/vcol/inc/vcol_supported_sql_funcs_main.inc 2009-10-16 22:57:48 +0000 +++ b/mysql-test/suite/vcol/inc/vcol_supported_sql_funcs_main.inc 2010-03-19 06:23:32 +0000 @@ -912,6 +912,7 @@ let $rows = 1; let $cols = a long, b datetime as (from_unixtime(a)); let $values1 = 1196440219,default; let $rows = 1; +set time_zone='UTC'; --source suite/vcol/inc/vcol_supported_sql_funcs.inc --echo # GET_FORMAT() === modified file 'mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result' --- a/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result 2010-03-17 02:32:31 +0000 +++ b/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result 2010-03-19 06:23:32 +0000 @@ -2194,6 +2194,7 @@ a b drop table t1; set sql_warnings = 0; # FROM_UNIXTIME() +set time_zone='UTC'; set sql_warnings = 1; create table t1 (a long, b datetime as (from_unixtime(a))); show create table t1; @@ -2205,7 +2206,7 @@ t1 CREATE TABLE `t1` ( insert into t1 values (1196440219,default); select * from t1; a b -1196440219 2007-11-30 08:30:19 +1196440219 2007-11-30 16:30:19 drop table t1; set sql_warnings = 0; # GET_FORMAT() === modified file 'mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result' --- a/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result 2010-03-17 02:32:31 +0000 +++ b/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result 2010-03-19 06:23:32 +0000 @@ -2194,6 +2194,7 @@ a b drop table t1; set sql_warnings = 0; # FROM_UNIXTIME() +set time_zone='UTC'; set sql_warnings = 1; create table t1 (a long, b datetime as (from_unixtime(a))); show create table t1; @@ -2205,7 +2206,7 @@ t1 CREATE TABLE `t1` ( insert into t1 values (1196440219,default); select * from t1; a b -1196440219 2007-11-30 08:30:19 +1196440219 2007-11-30 16:30:19 drop table t1; set sql_warnings = 0; # GET_FORMAT()

1 0

Re: [Maria-developers] Creating an external branch of sql-bench
by Kristian Nielsen 19 Mar '10

19 Mar '10

Hakan Kuecuekyilmaz <hakan(a)askmonty.org> writes: > we had some emails from Mark Callaghan, Patrick Galbraith, and Brian Aker > asking for an external branch of our sql-bench directory. They want to extend > and enhance the sql-bench benchmark suite. > > After a short phone call with Monty, he told me that there is a way to export > a directory structure with bzr in such a way, that merging it back to our > main tree would be easier. > > I checked http://askmonty.org/wiki/MariaDB::MergingMySQL but I could not find > a hint there. > > Monty told me, that we are doing something like that with the PBXT sources. > > Do you have an idea, what Monty meant? I know how to take an external project and merge it into MariaDB. This uses the "merge-into" bzr plugin. However, I do not know how to go in the opposite direction, that is take a part od the MariaDB tree and split it out in a separate project, and still allow merging it back in. Unless we want to delete it from our tree, re-create as a separate project, and merge that external project using merge-into. This would loose the revision history (I think). Also note that there seems to be a number of problems with this approach due to this bug: https://bugs.launchpad.net/bzr/+bug/375898 (For us, it works for xtradb, but not pbxt. Recently, some more people seem to also be hit by this, including MySQL people.) - Kristian.

1 0

[Maria-developers] Rev 2833: Fixed bug in view code when numeric reference in ORDER BY makes unusable View. in file:///Users/bell/maria/bzr/work-maria-5.1-view-order-bug/
by sanja＠askmonty.org 18 Mar '10

18 Mar '10

At file:///Users/bell/maria/bzr/work-maria-5.1-view-order-bug/ ------------------------------------------------------------ revno: 2833 revision-id: sanja(a)askmonty.org-20100318191914-wupwctzwixm1144h parent: sergii(a)pisem.net-20100312190521-jw1nggiv4427l5sm committer: sanja(a)askmonty.org branch nick: work-maria-5.1-view-order-bug timestamp: Thu 2010-03-18 21:19:14 +0200 message: Fixed bug in view code when numeric reference in ORDER BY makes unusable View. In view representation we prints expresion instead of its numeric reference. === modified file 'mysql-test/r/view.result' --- a/mysql-test/r/view.result 2010-02-10 19:06:24 +0000 +++ b/mysql-test/r/view.result 2010-03-18 19:19:14 +0000 @@ -3844,6 +3844,53 @@ ALTER TABLE v1; DROP VIEW v1; DROP TABLE t1; +# +# Maria Bug #???: ORDER BY column reference in view leads +# to unusable view +# +create table t1 (a int, b int); +insert into t1 values (2,70), (8, 30), (1, 20); +create view v1 as select a, b from t1 order by 2; +select v1.a from v1; +a +1 +8 +2 +show create view v1; +View Create View character_set_client collation_connection +v1 CREATE ALGORITHM=UNDEFINED DEFINER=`root`@`localhost` SQL SECURITY DEFINER VIEW `v1` AS select `t1`.`a` AS `a`,`t1`.`b` AS `b` from `t1` order by `t1`.`b` latin1 latin1_swedish_ci +drop view v1; +create view v1 as select a, b+3 as d from t1 order by 2; +select v1.a from v1; +a +1 +8 +2 +show create view v1; +View Create View character_set_client collation_connection +v1 CREATE ALGORITHM=UNDEFINED DEFINER=`root`@`localhost` SQL SECURITY DEFINER VIEW `v1` AS select `t1`.`a` AS `a`,(`t1`.`b` + 3) AS `d` from `t1` order by (`t1`.`b` + 3) latin1 latin1_swedish_ci +drop view v1; +create view v1 (a,v) as select a, b+3 as d from t1 order by 2; +select v1.a from v1; +a +1 +8 +2 +show create view v1; +View Create View character_set_client collation_connection +v1 CREATE ALGORITHM=UNDEFINED DEFINER=`root`@`localhost` SQL SECURITY DEFINER VIEW `v1` AS select `t1`.`a` AS `a`,(`t1`.`b` + 3) AS `v` from `t1` order by (`t1`.`b` + 3) latin1 latin1_swedish_ci +drop view v1; +create view v1 as select a, 3 as d from t1 order by 2; +select v1.a from v1; +a +2 +8 +1 +show create view v1; +View Create View character_set_client collation_connection +v1 CREATE ALGORITHM=UNDEFINED DEFINER=`root`@`localhost` SQL SECURITY DEFINER VIEW `v1` AS select `t1`.`a` AS `a`,3 AS `d` from `t1` order by (2 + 0) latin1 latin1_swedish_ci +drop view v1; +drop table t1; # ----------------------------------------------------------------- # -- End of 5.1 tests. # ----------------------------------------------------------------- === modified file 'mysql-test/t/view.test' --- a/mysql-test/t/view.test 2010-02-10 19:06:24 +0000 +++ b/mysql-test/t/view.test 2010-03-18 19:19:14 +0000 @@ -3869,6 +3869,29 @@ DROP VIEW v1; DROP TABLE t1; +--echo # +--echo # Maria Bug #???: ORDER BY column reference in view leads +--echo # to unusable view +--echo # +create table t1 (a int, b int); +insert into t1 values (2,70), (8, 30), (1, 20); +create view v1 as select a, b from t1 order by 2; +select v1.a from v1; +show create view v1; +drop view v1; +create view v1 as select a, b+3 as d from t1 order by 2; +select v1.a from v1; +show create view v1; +drop view v1; +create view v1 (a,v) as select a, b+3 as d from t1 order by 2; +select v1.a from v1; +show create view v1; +drop view v1; +create view v1 as select a, 3 as d from t1 order by 2; +select v1.a from v1; +show create view v1; +drop view v1; +drop table t1; --echo # ----------------------------------------------------------------- --echo # -- End of 5.1 tests. === modified file 'sql/item.cc' --- a/sql/item.cc 2010-03-04 08:03:07 +0000 +++ b/sql/item.cc 2010-03-18 19:19:14 +0000 @@ -2410,7 +2410,7 @@ void Item_string::print(String *str, enum_query_type query_type) { - if (query_type == QT_ORDINARY && is_cs_specified()) + if (query_type != QT_IS && is_cs_specified()) { str->append('_'); str->append(collation.collation->csname); @@ -2418,7 +2418,7 @@ str->append('\''); - if (query_type == QT_ORDINARY || + if (query_type != QT_IS || my_charset_same(str_value.charset(), system_charset_info)) { str_value.print(str); === modified file 'sql/mysql_priv.h' --- a/sql/mysql_priv.h 2010-03-04 08:03:07 +0000 +++ b/sql/mysql_priv.h 2010-03-18 19:19:14 +0000 @@ -52,12 +52,15 @@ QT_ORDINARY -- ordinary SQL query. QT_IS -- SQL query to be shown in INFORMATION_SCHEMA (in utf8 and without + QT_VIEW_INTERNAL -- view internal representation (like QT_ORDINARY except + ORDER BY clause) character set introducers). */ enum enum_query_type { QT_ORDINARY, - QT_IS + QT_IS, + QT_VIEW_INTERNAL }; /* TODO convert all these three maps to Bitmap classes */ === modified file 'sql/sql_lex.cc' --- a/sql/sql_lex.cc 2010-03-10 10:32:14 +0000 +++ b/sql/sql_lex.cc 2010-03-18 19:19:14 +0000 @@ -2057,9 +2057,27 @@ { if (order->counter_used) { - char buffer[20]; - size_t length= my_snprintf(buffer, 20, "%d", order->counter); - str->append(buffer, (uint) length); + if (query_type != QT_VIEW_INTERNAL) + { + char buffer[20]; + size_t length= my_snprintf(buffer, 20, "%d", order->counter); + str->append(buffer, (uint) length); + } + else + { + /* replace numeric reference with expression */ + if (order->item[0]->type() == Item::INT_ITEM && + order->item[0]->basic_const_item()) + { + char buffer[20]; + size_t length= my_snprintf(buffer, 20, "%d", order->counter); + str->append(buffer, (uint) length); + /* make it expression instead of integer constant */ + str->append(STRING_WITH_LEN("+0")); + } + else + (*order->item)->print(str, query_type); + } } else (*order->item)->print(str, query_type); @@ -2069,7 +2087,7 @@ str->append(','); } } - + void st_select_lex::print_limit(THD *thd, String *str, === modified file 'sql/sql_view.cc' --- a/sql/sql_view.cc 2010-03-04 08:03:07 +0000 +++ b/sql/sql_view.cc 2010-03-18 19:19:14 +0000 @@ -814,7 +814,7 @@ ulong sql_mode= thd->variables.sql_mode & MODE_ANSI_QUOTES; thd->variables.sql_mode&= ~MODE_ANSI_QUOTES; - lex->unit.print(&view_query, QT_ORDINARY); + lex->unit.print(&view_query, QT_VIEW_INTERNAL); lex->unit.print(&is_query, QT_IS); thd->variables.sql_mode|= sql_mode;

1 0

[Maria-developers] bzr commit into Mariadb 5.2, with Maria 2.0:maria/5.2 branch (knielsen:2745)
by knielsen＠knielsen-hq.org 18 Mar '10

18 Mar '10

#At lp:maria/5.2 2745 knielsen(a)knielsen-hq.org 2010-03-18 Fix merge errors in configure.in Add vcol test suite to `make dist`. modified: configure.in mysql-test/Makefile.am === modified file 'configure.in' --- a/configure.in 2010-03-15 11:51:23 +0000 +++ b/configure.in 2010-03-18 12:08:39 +0000 @@ -4,9 +4,6 @@ dnl Process this file with autoconf to p # Minimum Autoconf version required. AC_PREREQ(2.59) -# Minimum Autoconf version required. -AC_PREREQ(2.59) - # Remember to also update version.c in ndb. # When changing major version number please also check switch statement # in client/mysqlbinlog.cc / check_master_version(). @@ -15,7 +12,7 @@ AC_PREREQ(2.59) # MySQL version number. # # Note: the following line must be parseable by win/configure.js:GetVersion() -AC_INIT([MariaDB Server], [5.2.0-MariaDB], [], [mysql]) +AC_INIT([MariaDB Server], [5.2.0-MariaDB-alpha], [], [mysql]) AC_CONFIG_SRCDIR([sql/mysqld.cc]) AC_CANONICAL_SYSTEM # USTAR format gives us the possibility to store longer path names in === modified file 'mysql-test/Makefile.am' --- a/mysql-test/Makefile.am 2009-12-03 11:19:05 +0000 +++ b/mysql-test/Makefile.am 2010-03-18 12:08:39 +0000 @@ -102,7 +102,8 @@ TEST_DIRS = t r include std_data std_dat suite/rpl_ndb suite/rpl_ndb/t suite/rpl_ndb/r \ suite/parts suite/parts/t suite/parts/r suite/parts/inc \ suite/pbxt/t suite/pbxt/r \ - suite/innodb suite/innodb/t suite/innodb/r suite/innodb/include + suite/innodb suite/innodb/t suite/innodb/r suite/innodb/include \ + suite/vcol suite/vcol/t suite/vcol/r suite/vcol/inc # Used by dist-hook and install-data-local to copy all # test files into either dist or install directory

1 0

[Maria-developers] bzr commit into Mariadb 5.2, with Maria 2.0:maria/5.2 branch (knielsen:2745)
by knielsen＠knielsen-hq.org 18 Mar '10

18 Mar '10

#At lp:maria/5.2 2745 knielsen(a)knielsen-hq.org 2010-03-18 Fix merge errors in configure.in modified: configure.in === modified file 'configure.in' --- a/configure.in 2010-03-15 11:51:23 +0000 +++ b/configure.in 2010-03-18 11:51:40 +0000 @@ -4,9 +4,6 @@ dnl Process this file with autoconf to p # Minimum Autoconf version required. AC_PREREQ(2.59) -# Minimum Autoconf version required. -AC_PREREQ(2.59) - # Remember to also update version.c in ndb. # When changing major version number please also check switch statement # in client/mysqlbinlog.cc / check_master_version(). @@ -15,7 +12,7 @@ AC_PREREQ(2.59) # MySQL version number. # # Note: the following line must be parseable by win/configure.js:GetVersion() -AC_INIT([MariaDB Server], [5.2.0-MariaDB], [], [mysql]) +AC_INIT([MariaDB Server], [5.2.0-MariaDB-alpha], [], [mysql]) AC_CONFIG_SRCDIR([sql/mysqld.cc]) AC_CANONICAL_SYSTEM # USTAR format gives us the possibility to store longer path names in

1 0

[Maria-developers] WL enhancements
by Sergei Golubchik 17 Mar '10

17 Mar '10

Hi. That's a rough list of enhancement ideas for worklog. Some of them I'm going to do (but not all). Feel free to suggest more or "get rid of WL, use X instead", just don't forget the "because of" part. report: easy to do tasks make it to provide a list of "low-handing fruits", tasks that could be done relatively easily and without prior extensive MySQL source knowledge. For community members that want to help. report: who's doing what NOW that's pretty obvious report: tasks for specific version, roadmap worklog kind of does it now, mostly listed for completeness ability to remove hours Somebody mentioned that a number of hours could be increased by mistake, and there should be a way to decrease is back. I'm not convinced it's a good idea, though. generate weekly report templates to reduce the need for double or tripple reporting. WL generates a weekly report for a developer, based on its data, and sends it to this developer. The developer in question can edit it and sent to reports@, or copy-paste from it, or filter it out and ignore completely. private tasks and categories There should be a way to have private categories and tasks in WL. this includes fixing "private" field in WL tasks. distingushing between employees and users for many tasks from the above WL needs to distinguish between MPAB employees and other registered users. better search Sanja seems to be unhappy with WL search make it more readable for novice wl readers no big redesign or anything, though. but I think that moving task description up and, say, attached files and estimated number of hours down could help somewhat. not editable unless authenticated that's more a bug than a new feature. no changes in tasks should be allowed unless a user is authenticated. subscribe w/o authentication but perhaps we may want to allow users to subscribe to tasks w/o being authenticated ? I'm not going to do that, though. embeddable views a couple of pages that could be easily inserted (iframe-ed ?) into other pages without disrupting the design too much. Mainly "roadmap" and "easy tasks" reports, but also "WL of the day" too. "WL of the day" - it's a new crazy idea, I wanted to discuss. Basically, that's a small block somewhere on the main page that shows a randomly (?) selected WL task - only the description (or a part of it), a link to a full task page, and voting controls (useless... very important). Regards, Sergei

6 10

[Maria-developers] bzr commit into Mariadb 5.2, with Maria 2.0:maria/5.2 branch (igor:2744) Bug#539643
by Igor Babaev 17 Mar '10

17 Mar '10

#At lp:maria/5.2 based on revid:sergii@pisem.net-20100315115123-21tgprclhz7qbk6m 2744 Igor Babaev 2010-03-16 Fixed bug #539643. The cause of the problem is a bad merge MariaDB-5.1=>MariaDB-5.2. Added the vcol suite to the list of the default suites run by mysql-test-run.pl. modified: mysql-test/mysql-test-run.pl mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result sql/field.cc === modified file 'mysql-test/mysql-test-run.pl' --- a/mysql-test/mysql-test-run.pl 2010-03-10 10:32:14 +0000 +++ b/mysql-test/mysql-test-run.pl 2010-03-17 02:32:31 +0000 @@ -126,7 +126,7 @@ my $path_config_file; # The ge # executables will be used by the test suite. our $opt_vs_config = $ENV{'MTR_VS_CONFIG'}; -my $DEFAULT_SUITES= "main,binlog,federated,rpl,maria,parts"; +my $DEFAULT_SUITES= "main,binlog,federated,rpl,maria,parts,vcol"; my $opt_suites; our $opt_verbose= 0; # Verbose output, enable with --verbose === modified file 'mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result' --- a/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result 2009-10-16 22:57:48 +0000 +++ b/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_innodb.result 2010-03-17 02:32:31 +0000 @@ -2205,7 +2205,7 @@ t1 CREATE TABLE `t1` ( insert into t1 values (1196440219,default); select * from t1; a b -1196440219 2007-11-30 19:30:19 +1196440219 2007-11-30 08:30:19 drop table t1; set sql_warnings = 0; # GET_FORMAT() === modified file 'mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result' --- a/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result 2009-10-16 22:57:48 +0000 +++ b/mysql-test/suite/vcol/r/vcol_supported_sql_funcs_myisam.result 2010-03-17 02:32:31 +0000 @@ -2205,7 +2205,7 @@ t1 CREATE TABLE `t1` ( insert into t1 values (1196440219,default); select * from t1; a b -1196440219 2007-11-30 19:30:19 +1196440219 2007-11-30 08:30:19 drop table t1; set sql_warnings = 0; # GET_FORMAT() === modified file 'sql/field.cc' --- a/sql/field.cc 2010-03-15 11:51:23 +0000 +++ b/sql/field.cc 2010-03-17 02:32:31 +0000 @@ -9598,13 +9598,13 @@ bool Create_field::init(THD *thd, char * interval_list.empty(); comment= *fld_comment; + vcol_info= fld_vcol_info; stored_in_db= TRUE; /* Initialize data for a computed field */ if ((uchar)fld_type == (uchar)MYSQL_TYPE_VIRTUAL) { DBUG_ASSERT(vcol_info && vcol_info->expr_item); - vcol_info= fld_vcol_info; stored_in_db= vcol_info->is_stored(); /* Walk through the Item tree checking if all items are valid @@ -9624,8 +9624,6 @@ bool Create_field::init(THD *thd, char * */ sql_type= fld_type= vcol_info->get_real_type(); } - else - vcol_info= NULL; /* Set NO_DEFAULT_VALUE_FLAG if this field doesn't have a default value and

1 0

[Maria-developers] Updated (by Serg): Phone home (12)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Phone home CREATION DATE..: Wed, 01 Apr 2009, 15:30 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Bothorsen COPIES TO......: Sergei CATEGORY.......: Server-RawIdeaBin TASK ID........: 12 (http://askmonty.org/worklog/?tid=12) VERSION........: Connector/.NET-1.6 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 120 (hours remain) ORIG. ESTIMATE.: 120 PROGRESS NOTES: -=-=(Serg - Tue, 16 Mar 2010, 19:51)=-=- Observers changed: Sergei -=-=(Monty - Tue, 09 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.12.old.27527 2010-03-09 19:28:51.000000000 +0000 +++ /tmp/wklog.12.new.27527 2010-03-09 19:28:51.000000000 +0000 @@ -2,10 +2,10 @@ a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and -analysis of this information will give MySQL AB useful insight into +analysis of this information will give MariaDB developers useful insight into the user base -Summary of collected information +Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on @@ -28,7 +28,7 @@ gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name - (optionally by user) MySQL customer support contract id + (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database -=-=(Monty - Tue, 09 Mar 2010, 19:25)=-=- High Level Description modified. --- /tmp/wklog.12.old.27502 2010-03-09 19:25:56.000000000 +0000 +++ /tmp/wklog.12.new.27502 2010-03-09 19:25:56.000000000 +0000 @@ -1,6 +1,6 @@ -This project is to develop a plugin for the MySQL server that provides +This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database -server will regularly contact a web service operated by MySQL AB, and +server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MySQL AB useful insight into the user base @@ -32,10 +32,10 @@ Data Not Sent Contents or names of any user database - Anything that allows MySQL to track down the user + Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it -What will run at MySQL's datacenter +What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database DESCRIPTION: This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MariaDB developers useful insight into the user base Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on what version and build they are running what features are being used Information sent from each instance A unique server identifier secure hash of MAC address + listening port unique, but doesn't leak customer data Processor type, speed, processor count / core count, bitwidth (32/64) OS / Distro / Kernel id and version Which storage engines are in use Number and size of databases (disk space, probably can't for cluster) Counts / Rates of I/O activity List of loaded plugins SHOW STATUS SHOW VARIABLES (but not anything that can give away user identity) will be a explicit list of what variables will be shown gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database database schema is TBD, but not complicated Analysis/Reporting/Business Process are TBD What will run on the users' machines Daemon Plugin module can be dynamically loaded, or statically compiled runs as part of the mysqld process daemon plugins have full access to server internals Start and use it's own thread will not block normal operation will yield often will hold read mutexs as short as possible Loop and delay on an interval (specified by option) probably default to be on server restart and about once a week randomly spread time to avoid too many calling in at the same moment Gather data from internal mysqld data structures Convert data into simple text format (human readable) Transmit data via HTTP REST POST to one or more given URLs ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Serg): Phone home (12)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Phone home CREATION DATE..: Wed, 01 Apr 2009, 15:30 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Bothorsen COPIES TO......: Sergei CATEGORY.......: Server-RawIdeaBin TASK ID........: 12 (http://askmonty.org/worklog/?tid=12) VERSION........: Connector/.NET-1.6 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 120 (hours remain) ORIG. ESTIMATE.: 120 PROGRESS NOTES: -=-=(Serg - Tue, 16 Mar 2010, 19:51)=-=- Observers changed: Sergei -=-=(Monty - Tue, 09 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.12.old.27527 2010-03-09 19:28:51.000000000 +0000 +++ /tmp/wklog.12.new.27527 2010-03-09 19:28:51.000000000 +0000 @@ -2,10 +2,10 @@ a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and -analysis of this information will give MySQL AB useful insight into +analysis of this information will give MariaDB developers useful insight into the user base -Summary of collected information +Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on @@ -28,7 +28,7 @@ gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name - (optionally by user) MySQL customer support contract id + (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database -=-=(Monty - Tue, 09 Mar 2010, 19:25)=-=- High Level Description modified. --- /tmp/wklog.12.old.27502 2010-03-09 19:25:56.000000000 +0000 +++ /tmp/wklog.12.new.27502 2010-03-09 19:25:56.000000000 +0000 @@ -1,6 +1,6 @@ -This project is to develop a plugin for the MySQL server that provides +This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database -server will regularly contact a web service operated by MySQL AB, and +server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MySQL AB useful insight into the user base @@ -32,10 +32,10 @@ Data Not Sent Contents or names of any user database - Anything that allows MySQL to track down the user + Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it -What will run at MySQL's datacenter +What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database DESCRIPTION: This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MariaDB developers useful insight into the user base Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on what version and build they are running what features are being used Information sent from each instance A unique server identifier secure hash of MAC address + listening port unique, but doesn't leak customer data Processor type, speed, processor count / core count, bitwidth (32/64) OS / Distro / Kernel id and version Which storage engines are in use Number and size of databases (disk space, probably can't for cluster) Counts / Rates of I/O activity List of loaded plugins SHOW STATUS SHOW VARIABLES (but not anything that can give away user identity) will be a explicit list of what variables will be shown gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database database schema is TBD, but not complicated Analysis/Reporting/Business Process are TBD What will run on the users' machines Daemon Plugin module can be dynamically loaded, or statically compiled runs as part of the mysqld process daemon plugins have full access to server internals Start and use it's own thread will not block normal operation will yield often will hold read mutexs as short as possible Loop and delay on an interval (specified by option) probably default to be on server restart and about once a week randomly spread time to avoid too many calling in at the same moment Gather data from internal mysqld data structures Convert data into simple text format (human readable) Transmit data via HTTP REST POST to one or more given URLs ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Serg): Phone home (12)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Phone home CREATION DATE..: Wed, 01 Apr 2009, 15:30 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Bothorsen COPIES TO......: Sergei CATEGORY.......: Server-RawIdeaBin TASK ID........: 12 (http://askmonty.org/worklog/?tid=12) VERSION........: Connector/.NET-1.6 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 120 (hours remain) ORIG. ESTIMATE.: 120 PROGRESS NOTES: -=-=(Serg - Tue, 16 Mar 2010, 19:51)=-=- Observers changed: Sergei -=-=(Monty - Tue, 09 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.12.old.27527 2010-03-09 19:28:51.000000000 +0000 +++ /tmp/wklog.12.new.27527 2010-03-09 19:28:51.000000000 +0000 @@ -2,10 +2,10 @@ a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and -analysis of this information will give MySQL AB useful insight into +analysis of this information will give MariaDB developers useful insight into the user base -Summary of collected information +Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on @@ -28,7 +28,7 @@ gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name - (optionally by user) MySQL customer support contract id + (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database -=-=(Monty - Tue, 09 Mar 2010, 19:25)=-=- High Level Description modified. --- /tmp/wklog.12.old.27502 2010-03-09 19:25:56.000000000 +0000 +++ /tmp/wklog.12.new.27502 2010-03-09 19:25:56.000000000 +0000 @@ -1,6 +1,6 @@ -This project is to develop a plugin for the MySQL server that provides +This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database -server will regularly contact a web service operated by MySQL AB, and +server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MySQL AB useful insight into the user base @@ -32,10 +32,10 @@ Data Not Sent Contents or names of any user database - Anything that allows MySQL to track down the user + Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it -What will run at MySQL's datacenter +What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database DESCRIPTION: This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MariaDB developers useful insight into the user base Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on what version and build they are running what features are being used Information sent from each instance A unique server identifier secure hash of MAC address + listening port unique, but doesn't leak customer data Processor type, speed, processor count / core count, bitwidth (32/64) OS / Distro / Kernel id and version Which storage engines are in use Number and size of databases (disk space, probably can't for cluster) Counts / Rates of I/O activity List of loaded plugins SHOW STATUS SHOW VARIABLES (but not anything that can give away user identity) will be a explicit list of what variables will be shown gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database database schema is TBD, but not complicated Analysis/Reporting/Business Process are TBD What will run on the users' machines Daemon Plugin module can be dynamically loaded, or statically compiled runs as part of the mysqld process daemon plugins have full access to server internals Start and use it's own thread will not block normal operation will yield often will hold read mutexs as short as possible Loop and delay on an interval (specified by option) probably default to be on server restart and about once a week randomly spread time to avoid too many calling in at the same moment Gather data from internal mysqld data structures Convert data into simple text format (human readable) Transmit data via HTTP REST POST to one or more given URLs ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Serg): Phone home (12)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Phone home CREATION DATE..: Wed, 01 Apr 2009, 15:30 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Bothorsen COPIES TO......: Sergei CATEGORY.......: Server-RawIdeaBin TASK ID........: 12 (http://askmonty.org/worklog/?tid=12) VERSION........: Connector/.NET-1.6 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 120 (hours remain) ORIG. ESTIMATE.: 120 PROGRESS NOTES: -=-=(Serg - Tue, 16 Mar 2010, 19:51)=-=- Observers changed: Sergei -=-=(Monty - Tue, 09 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.12.old.27527 2010-03-09 19:28:51.000000000 +0000 +++ /tmp/wklog.12.new.27527 2010-03-09 19:28:51.000000000 +0000 @@ -2,10 +2,10 @@ a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and -analysis of this information will give MySQL AB useful insight into +analysis of this information will give MariaDB developers useful insight into the user base -Summary of collected information +Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on @@ -28,7 +28,7 @@ gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name - (optionally by user) MySQL customer support contract id + (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database -=-=(Monty - Tue, 09 Mar 2010, 19:25)=-=- High Level Description modified. --- /tmp/wklog.12.old.27502 2010-03-09 19:25:56.000000000 +0000 +++ /tmp/wklog.12.new.27502 2010-03-09 19:25:56.000000000 +0000 @@ -1,6 +1,6 @@ -This project is to develop a plugin for the MySQL server that provides +This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database -server will regularly contact a web service operated by MySQL AB, and +server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MySQL AB useful insight into the user base @@ -32,10 +32,10 @@ Data Not Sent Contents or names of any user database - Anything that allows MySQL to track down the user + Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it -What will run at MySQL's datacenter +What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database DESCRIPTION: This project is to develop a plugin for the MariaDB server that provides a "phone home" feature. When this plugin is installed, the database server will regularly contact a web service operated by Monty Program Ab, and upload a bundle of non-sensitive information. The collection and analysis of this information will give MariaDB developers useful insight into the user base Summary of collected information (that anyone will be allowed to access) how many servers are running the plugin - can help estimate total number of running servers worldwide what platform & hardware they are running on what version and build they are running what features are being used Information sent from each instance A unique server identifier secure hash of MAC address + listening port unique, but doesn't leak customer data Processor type, speed, processor count / core count, bitwidth (32/64) OS / Distro / Kernel id and version Which storage engines are in use Number and size of databases (disk space, probably can't for cluster) Counts / Rates of I/O activity List of loaded plugins SHOW STATUS SHOW VARIABLES (but not anything that can give away user identity) will be a explicit list of what variables will be shown gives among other things, the MySQL server version & Server uptime (optionally by user) geographic location (optionally by user) user information / company name (optionally by user) Monty Program Ab customer support contract id Data Not Sent Contents or names of any user database Anything that allows MariaDB to track down the user if the user doesn't explicitly permit it What will run at Monty Program's or/and the users datacenter simple CGI on Apache takes a HTTP REST PUT insert received information into a database database schema is TBD, but not complicated Analysis/Reporting/Business Process are TBD What will run on the users' machines Daemon Plugin module can be dynamically loaded, or statically compiled runs as part of the mysqld process daemon plugins have full access to server internals Start and use it's own thread will not block normal operation will yield often will hold read mutexs as short as possible Loop and delay on an interval (specified by option) probably default to be on server restart and about once a week randomly spread time to avoid too many calling in at the same moment Gather data from internal mysqld data structures Convert data into simple text format (human readable) Transmit data via HTTP REST POST to one or more given URLs ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (85)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:10 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 85 (http://askmonty.org/worklog/?tid=85) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:34)=-=- High Level Description modified. --- /tmp/wklog.85.old.22371 2010-03-16 19:34:33.000000000 +0000 +++ /tmp/wklog.85.new.22371 2010-03-16 19:34:33.000000000 +0000 @@ -15,4 +15,5 @@ the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of -our external contributers. +our external contributers (see the attached file segmented_keycache_v2.diff with +the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Category updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Version updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 -=-=(Igor - Sun, 14 Feb 2010, 00:12)=-=- New attachment: 'segmented_keycache_v2.diff' DESCRIPTION: A partitioned key cache is a collection of structures for regular MyiSAM key caches called key cache partitions. Any page from a file can be placed into a buffer of only one partition. The number of the partition is calculated from the file number and the position of the page in the file, and it's always the same for the page. The function that maps pages into partitions takes care of even distribution of pages among partitions. Partition key cache mitigate one of the major problem of simple key cache: thread contention for key cache lock (mutex). Every call of a key cache interface function must acquire this lock. So threads compete for this lock even in the case when they have acquired shared locks for the file and pages they want read from are in the key cache buffers. When working with a partitioned key cache any key cache interface function that needs only one page has to acquire the key cache lock only for the partition the page is ascribed to. This makes the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of our external contributers (see the attached file segmented_keycache_v2.diff with the original patch from the contributor). ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (85)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:10 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 85 (http://askmonty.org/worklog/?tid=85) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:34)=-=- High Level Description modified. --- /tmp/wklog.85.old.22371 2010-03-16 19:34:33.000000000 +0000 +++ /tmp/wklog.85.new.22371 2010-03-16 19:34:33.000000000 +0000 @@ -15,4 +15,5 @@ the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of -our external contributers. +our external contributers (see the attached file segmented_keycache_v2.diff with +the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Category updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Version updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 -=-=(Igor - Sun, 14 Feb 2010, 00:12)=-=- New attachment: 'segmented_keycache_v2.diff' DESCRIPTION: A partitioned key cache is a collection of structures for regular MyiSAM key caches called key cache partitions. Any page from a file can be placed into a buffer of only one partition. The number of the partition is calculated from the file number and the position of the page in the file, and it's always the same for the page. The function that maps pages into partitions takes care of even distribution of pages among partitions. Partition key cache mitigate one of the major problem of simple key cache: thread contention for key cache lock (mutex). Every call of a key cache interface function must acquire this lock. So threads compete for this lock even in the case when they have acquired shared locks for the file and pages they want read from are in the key cache buffers. When working with a partitioned key cache any key cache interface function that needs only one page has to acquire the key cache lock only for the partition the page is ascribed to. This makes the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of our external contributers (see the attached file segmented_keycache_v2.diff with the original patch from the contributor). ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (85)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:10 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 85 (http://askmonty.org/worklog/?tid=85) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:34)=-=- High Level Description modified. --- /tmp/wklog.85.old.22371 2010-03-16 19:34:33.000000000 +0000 +++ /tmp/wklog.85.new.22371 2010-03-16 19:34:33.000000000 +0000 @@ -15,4 +15,5 @@ the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of -our external contributers. +our external contributers (see the attached file segmented_keycache_v2.diff with +the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Category updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:15)=-=- Version updated. --- /tmp/wklog.85.old.9810 2010-02-13 22:15:43.000000000 +0000 +++ /tmp/wklog.85.new.9810 2010-02-13 22:15:43.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 -=-=(Igor - Sun, 14 Feb 2010, 00:12)=-=- New attachment: 'segmented_keycache_v2.diff' DESCRIPTION: A partitioned key cache is a collection of structures for regular MyiSAM key caches called key cache partitions. Any page from a file can be placed into a buffer of only one partition. The number of the partition is calculated from the file number and the position of the page in the file, and it's always the same for the page. The function that maps pages into partitions takes care of even distribution of pages among partitions. Partition key cache mitigate one of the major problem of simple key cache: thread contention for key cache lock (mutex). Every call of a key cache interface function must acquire this lock. So threads compete for this lock even in the case when they have acquired shared locks for the file and pages they want read from are in the key cache buffers. When working with a partitioned key cache any key cache interface function that needs only one page has to acquire the key cache lock only for the partition the page is ascribed to. This makes the chances for threads not compete for the same key cache lock better. The idea and the original of the partitioned key cache was provided by one of our external contributers (see the attached file segmented_keycache_v2.diff with the original patch from the contributor). ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:30)=-=- Title modified. --- /tmp/wklog.86.old.22309 2010-03-16 19:30:04.000000000 +0000 +++ /tmp/wklog.86.new.22309 2010-03-16 19:30:04.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:30)=-=- Title modified. --- /tmp/wklog.86.old.22309 2010-03-16 19:30:04.000000000 +0000 +++ /tmp/wklog.86.new.22309 2010-03-16 19:30:04.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:30)=-=- Title modified. --- /tmp/wklog.86.old.22309 2010-03-16 19:30:04.000000000 +0000 +++ /tmp/wklog.86.new.22309 2010-03-16 19:30:04.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (86)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:17 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-Sprint TASK ID........: 86 (http://askmonty.org/worklog/?tid=86) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 80 (hours remain) ORIG. ESTIMATE.: 80 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:29)=-=- High Level Description modified. --- /tmp/wklog.86.old.22292 2010-03-16 19:29:37.000000000 +0000 +++ /tmp/wklog.86.new.22292 2010-03-16 19:29:37.000000000 +0000 @@ -1,19 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers (see the attached file segmented_keycache_v2.diff with -the original patch from the contributor). -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Privacy level updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -y +n -=-=(Igor - Sun, 14 Feb 2010, 00:19)=-=- Category updated. --- /tmp/wklog.86.old.10092 2010-02-13 22:19:03.000000000 +0000 +++ /tmp/wklog.86.new.10092 2010-02-13 22:19:03.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Server-Sprint -=-=(Igor - Sun, 14 Feb 2010, 00:18)=-=- Version updated. --- /tmp/wklog.86.old.10044 2010-02-14 00:18:31.000000000 +0200 +++ /tmp/wklog.86.new.10044 2010-02-14 00:18:31.000000000 +0200 @@ -1 +1 @@ -Benchmarks-3.0 +Server-5.2 DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Title modified. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Version updated. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-9.x -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Title modified. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Version updated. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-9.x -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Unused (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Unused CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Title modified. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Partitioned Key Cache for MyISAM +Unused -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- Version updated. --- /tmp/wklog.84.old.22271 2010-03-16 19:28:50.000000000 +0000 +++ /tmp/wklog.84.new.22271 2010-03-16 19:28:50.000000000 +0000 @@ -1 +1 @@ -Benchmarks-3.0 +Server-9.x -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Benchmarks-3.0 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Benchmarks-3.0 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Partitioned Key Cache for MyISAM (84)
by worklog-noreply＠askmonty.org 16 Mar '10

16 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Partitioned Key Cache for MyISAM CREATION DATE..: Sun, 14 Feb 2010, 00:09 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Monty, Sergei CATEGORY.......: Server-BackLog TASK ID........: 84 (http://askmonty.org/worklog/?tid=84) VERSION........: Benchmarks-3.0 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Tue, 16 Mar 2010, 19:28)=-=- High Level Description modified. --- /tmp/wklog.84.old.22253 2010-03-16 19:28:09.000000000 +0000 +++ /tmp/wklog.84.new.22253 2010-03-16 19:28:09.000000000 +0000 @@ -1,18 +1 @@ -A partitioned key cache is a collection of structures for regular MyiSAM key -caches called key cache partitions. Any page from a file can be placed into a -buffer of only one partition. The number of the partition is calculated from the -file number and the position of the page in the file, and it's always the same -for the page. The function that maps pages into partitions takes care of even -distribution of pages among partitions. -Partition key cache mitigate one of the major problem of simple key cache: -thread contention for key cache lock (mutex). Every call of a key cache -interface function must acquire this lock. So threads compete for this lock even -in the case when they have acquired shared locks for the file and pages they -want read from are in the key cache buffers. When working with a partitioned key -cache any key cache interface function that needs only one page has to acquire -the key cache lock only for the partition the page is ascribed to. This makes -the chances for threads not compete for the same key cache lock better. - -The idea and the original of the partitioned key cache was provided by one of -our external contributers. DESCRIPTION: ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Rev 2780: MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs in file:///home/tsk/mprog/src/5.3-subqueries/
by timour＠askmonty.org 15 Mar '10

15 Mar '10

At file:///home/tsk/mprog/src/5.3-subqueries/ ------------------------------------------------------------ revno: 2780 revision-id: timour(a)askmonty.org-20100315224130-321rym1lsuwz2j5z parent: timour(a)askmonty.org-20100315195258-nhomb3anbb1tv3mi committer: timour(a)askmonty.org branch nick: 5.3-subqueries timestamp: Tue 2010-03-16 00:41:30 +0200 message: MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs Fix for the PBXT copy of subselect.test. === modified file 'mysql-test/suite/pbxt/r/subselect.result' --- a/mysql-test/suite/pbxt/r/subselect.result 2010-02-23 09:22:02 +0000 +++ b/mysql-test/suite/pbxt/r/subselect.result 2010-03-15 22:41:30 +0000 @@ -876,6 +876,8 @@ 4.5 NULL drop table t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int(11) NOT NULL default '0', PRIMARY KEY (a)); CREATE TABLE t2 (a int(11) default '0', INDEX (a)); INSERT INTO t1 VALUES (1),(2),(3),(4); @@ -1771,6 +1773,7 @@ Warnings: Note 1003 select `test`.`a`.`id` AS `id`,`test`.`a`.`text` AS `text`,`test`.`b`.`id` AS `id`,`test`.`b`.`text` AS `text`,`test`.`c`.`id` AS `id`,`test`.`c`.`text` AS `text` from `test`.`t1` `a` left join `test`.`t2` `b` on(((`test`.`b`.`id` = `test`.`a`.`id`) or isnull(`test`.`b`.`id`))) join `test`.`t1` `c` where (if(isnull(`test`.`b`.`id`),1000,`test`.`b`.`id`) = `test`.`c`.`id`) drop table t1,t2; +set @@optimizer_switch=@save_optimizer_switch; create table t1 (a int); insert into t1 values (1); explain select benchmark(1000, (select a from t1 where a=sha(rand()))); @@ -2750,6 +2753,8 @@ max(fld) 1 drop table t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (one int, two int, flag char(1)); CREATE TABLE t2 (one int, two int, flag char(1)); INSERT INTO t1 VALUES(1,2,'Y'),(2,3,'Y'),(3,4,'Y'),(5,6,'N'),(7,8,'N'); @@ -2834,6 +2839,7 @@ Warnings: Note 1003 select `test`.`t1`.`one` AS `one`,`test`.`t1`.`two` AS `two`,<in_optimizer>((`test`.`t1`.`one`,`test`.`t1`.`two`),<exists>(select `test`.`t2`.`one` AS `one`,`test`.`t2`.`two` AS `two` from `test`.`t2` where (`test`.`t2`.`flag` = '0') group by `test`.`t2`.`one`,`test`.`t2`.`two` having (trigcond(((<cache>(`test`.`t1`.`one`) = `test`.`t2`.`one`) or isnull(`test`.`t2`.`one`))) and trigcond(((<cache>(`test`.`t1`.`two`) = `test`.`t2`.`two`) or isnull(`test`.`t2`.`two`))) and trigcond(<is_not_null_test>(`test`.`t2`.`one`)) and trigcond(<is_not_null_test>(`test`.`t2`.`two`))))) AS `test` from `test`.`t1` DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a char(5), b char(5)); INSERT INTO t1 VALUES (NULL,'aaa'), ('aaa','aaa'); SELECT * FROM t1 WHERE (a,b) IN (('aaa','aaa'), ('aaa','bbb')); @@ -3004,6 +3010,8 @@ 1 1 1 3 DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1(a int, INDEX (a)); INSERT INTO t1 VALUES (1), (3), (5), (7); INSERT INTO t1 VALUES (NULL); @@ -3019,6 +3027,7 @@ 2 NULL 3 1 DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a DATETIME); INSERT INTO t1 VALUES ('1998-09-23'), ('2003-03-25'); CREATE TABLE t2 AS SELECT === modified file 'mysql-test/suite/pbxt/t/subselect.test' --- a/mysql-test/suite/pbxt/t/subselect.test 2009-11-06 17:22:32 +0000 +++ b/mysql-test/suite/pbxt/t/subselect.test 2010-03-15 22:41:30 +0000 @@ -477,6 +477,9 @@ # Null with keys # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1 (a int(11) NOT NULL default '0', PRIMARY KEY (a)); CREATE TABLE t2 (a int(11) default '0', INDEX (a)); INSERT INTO t1 VALUES (1),(2),(3),(4); @@ -1121,6 +1124,8 @@ explain extended select * from t1 a left join t2 b on (a.id=b.id or b.id is null) join t1 c on (if(isnull(b.id), 1000, b.id)=c.id); drop table t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Static tables & rund() in subqueries # @@ -1784,6 +1789,9 @@ # Bug #11867: queries with ROW(,elems>) IN (SELECT DISTINCT <cols> FROM ...) # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1 (one int, two int, flag char(1)); CREATE TABLE t2 (one int, two int, flag char(1)); INSERT INTO t1 VALUES(1,2,'Y'),(2,3,'Y'),(3,4,'Y'),(5,6,'N'),(7,8,'N'); @@ -1811,6 +1819,9 @@ explain extended SELECT one,two,ROW(one,two) IN (SELECT one,two FROM t2 WHERE flag = '0' group by one,two) as 'test' from t1; DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + + # # Bug #12392: where cond with IN predicate for rows and NULL values in table # @@ -1972,6 +1983,9 @@ # with possible NULL values by index access from the outer query # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1(a int, INDEX (a)); INSERT INTO t1 VALUES (1), (3), (5), (7); INSERT INTO t1 VALUES (NULL); @@ -1984,6 +1998,8 @@ DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #11302: getObject() returns a String for a sub-query of type datetime # @@ -3096,6 +3112,7 @@ DROP TABLE t1,t2; + # # Bug #32400: Complex SELECT query returns correct result only on some # occasions

1 0

[Maria-developers] bzr commit into file:///home/tsk/mprog/src/5.3-subqueries/ branch (timour:2780)
by timour＠askmonty.org 15 Mar '10

15 Mar '10

#At file:///home/tsk/mprog/src/5.3-subqueries/ based on revid:timour@askmonty.org-20100315195258-nhomb3anbb1tv3mi 2780 timour(a)askmonty.org 2010-03-16 MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs Fix for the PBXT copy of subselect.test. modified: mysql-test/suite/pbxt/r/subselect.result mysql-test/suite/pbxt/t/subselect.test === modified file 'mysql-test/suite/pbxt/r/subselect.result' --- a/mysql-test/suite/pbxt/r/subselect.result 2010-02-23 09:22:02 +0000 +++ b/mysql-test/suite/pbxt/r/subselect.result 2010-03-15 22:41:30 +0000 @@ -876,6 +876,8 @@ select (select a+1) from t1; 4.5 NULL drop table t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int(11) NOT NULL default '0', PRIMARY KEY (a)); CREATE TABLE t2 (a int(11) default '0', INDEX (a)); INSERT INTO t1 VALUES (1),(2),(3),(4); @@ -1771,6 +1773,7 @@ id select_type table type possible_keys Warnings: Note 1003 select `test`.`a`.`id` AS `id`,`test`.`a`.`text` AS `text`,`test`.`b`.`id` AS `id`,`test`.`b`.`text` AS `text`,`test`.`c`.`id` AS `id`,`test`.`c`.`text` AS `text` from `test`.`t1` `a` left join `test`.`t2` `b` on(((`test`.`b`.`id` = `test`.`a`.`id`) or isnull(`test`.`b`.`id`))) join `test`.`t1` `c` where (if(isnull(`test`.`b`.`id`),1000,`test`.`b`.`id`) = `test`.`c`.`id`) drop table t1,t2; +set @@optimizer_switch=@save_optimizer_switch; create table t1 (a int); insert into t1 values (1); explain select benchmark(1000, (select a from t1 where a=sha(rand()))); @@ -2750,6 +2753,8 @@ select * from (select max(fld) from t1) max(fld) 1 drop table t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (one int, two int, flag char(1)); CREATE TABLE t2 (one int, two int, flag char(1)); INSERT INTO t1 VALUES(1,2,'Y'),(2,3,'Y'),(3,4,'Y'),(5,6,'N'),(7,8,'N'); @@ -2834,6 +2839,7 @@ id select_type table type possible_keys Warnings: Note 1003 select `test`.`t1`.`one` AS `one`,`test`.`t1`.`two` AS `two`,<in_optimizer>((`test`.`t1`.`one`,`test`.`t1`.`two`),<exists>(select `test`.`t2`.`one` AS `one`,`test`.`t2`.`two` AS `two` from `test`.`t2` where (`test`.`t2`.`flag` = '0') group by `test`.`t2`.`one`,`test`.`t2`.`two` having (trigcond(((<cache>(`test`.`t1`.`one`) = `test`.`t2`.`one`) or isnull(`test`.`t2`.`one`))) and trigcond(((<cache>(`test`.`t1`.`two`) = `test`.`t2`.`two`) or isnull(`test`.`t2`.`two`))) and trigcond(<is_not_null_test>(`test`.`t2`.`one`)) and trigcond(<is_not_null_test>(`test`.`t2`.`two`))))) AS `test` from `test`.`t1` DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a char(5), b char(5)); INSERT INTO t1 VALUES (NULL,'aaa'), ('aaa','aaa'); SELECT * FROM t1 WHERE (a,b) IN (('aaa','aaa'), ('aaa','bbb')); @@ -3004,6 +3010,8 @@ field1 field2 1 1 1 3 DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1(a int, INDEX (a)); INSERT INTO t1 VALUES (1), (3), (5), (7); INSERT INTO t1 VALUES (NULL); @@ -3019,6 +3027,7 @@ a a IN (SELECT a FROM t1) 2 NULL 3 1 DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a DATETIME); INSERT INTO t1 VALUES ('1998-09-23'), ('2003-03-25'); CREATE TABLE t2 AS SELECT === modified file 'mysql-test/suite/pbxt/t/subselect.test' --- a/mysql-test/suite/pbxt/t/subselect.test 2009-11-06 17:22:32 +0000 +++ b/mysql-test/suite/pbxt/t/subselect.test 2010-03-15 22:41:30 +0000 @@ -477,6 +477,9 @@ drop table t1; # Null with keys # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1 (a int(11) NOT NULL default '0', PRIMARY KEY (a)); CREATE TABLE t2 (a int(11) default '0', INDEX (a)); INSERT INTO t1 VALUES (1),(2),(3),(4); @@ -1121,6 +1124,8 @@ select * from t1 a left join t2 b on (a. explain extended select * from t1 a left join t2 b on (a.id=b.id or b.id is null) join t1 c on (if(isnull(b.id), 1000, b.id)=c.id); drop table t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Static tables & rund() in subqueries # @@ -1784,6 +1789,9 @@ drop table t1; # Bug #11867: queries with ROW(,elems>) IN (SELECT DISTINCT <cols> FROM ...) # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1 (one int, two int, flag char(1)); CREATE TABLE t2 (one int, two int, flag char(1)); INSERT INTO t1 VALUES(1,2,'Y'),(2,3,'Y'),(3,4,'Y'),(5,6,'N'),(7,8,'N'); @@ -1811,6 +1819,9 @@ explain extended SELECT one,two from t1 explain extended SELECT one,two,ROW(one,two) IN (SELECT one,two FROM t2 WHERE flag = '0' group by one,two) as 'test' from t1; DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + + # # Bug #12392: where cond with IN predicate for rows and NULL values in table # @@ -1972,6 +1983,9 @@ DROP TABLE t1, t2; # with possible NULL values by index access from the outer query # +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + CREATE TABLE t1(a int, INDEX (a)); INSERT INTO t1 VALUES (1), (3), (5), (7); INSERT INTO t1 VALUES (NULL); @@ -1984,6 +1998,8 @@ SELECT a, a IN (SELECT a FROM t1) FROM t DROP TABLE t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #11302: getObject() returns a String for a sub-query of type datetime # @@ -3096,6 +3112,7 @@ SELECT a,b FROM t1 WHERE b IN (SELECT a DROP TABLE t1,t2; + # # Bug #32400: Complex SELECT query returns correct result only on some # occasions

1 0

[Maria-developers] Rev 2779: Merge in MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs in file:///home/tsk/mprog/src/5.3-subqueries/
by timour＠askmonty.org 15 Mar '10

15 Mar '10

At file:///home/tsk/mprog/src/5.3-subqueries/ ------------------------------------------------------------ revno: 2779 [merge] revision-id: timour(a)askmonty.org-20100315195258-nhomb3anbb1tv3mi parent: psergey(a)askmonty.org-20100315063535-jsp4jgya6lfqt8e6 parent: timour(a)askmonty.org-20100311214331-kw8ng8aiy6h60vai committer: timour(a)askmonty.org branch nick: 5.3-subqueries timestamp: Mon 2010-03-15 21:52:58 +0200 message: Merge in MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs modified: mysql-test/include/mix1.inc sp1f-innodb_mysql.test-20060426055153-mgtahdmgajg7vffqbq4xrmkzbhvanlaz mysql-test/r/index_merge_myisam.result sp1f-index_merge_myisam.r-20060816114353-wd2664hjxwyjdvm4snup647av5fmxfln mysql-test/r/innodb_mysql.result sp1f-innodb_mysql.result-20060426055153-bychbbfnqtvmvrwccwhn24i6yi46uqjv mysql-test/r/myisam_mrr.result myisam_mrr.result-20091215071345-6wadxunod6vi8m48-1 mysql-test/r/ps.result sp1f-ps.result-20040405154119-efxzt5onloys45nfjak4gt44kr4awkdi mysql-test/r/subselect.result sp1f-subselect.result-20020512204640-zgegcsgavnfd7t7eyrf7ibuqomsw7uzo mysql-test/r/subselect3.result sp1f-subselect3.result-20061031174245-v7hvtc7uwevifiq4lziwv5gdcxpeak7t mysql-test/r/subselect3_jcl6.result subselect3_jcl6.resu-20100117143923-cf6j4mu5zzng00u7-1 mysql-test/r/subselect_no_mat.result subselect_no_mat.res-20100117143924-hut18sl9k2c7qdj8-1 mysql-test/r/subselect_no_opts.result subselect_no_opts.re-20100117143925-pabg7o8iyokjlu93-1 mysql-test/r/subselect_no_semijoin.result subselect_no_semijoi-20100117143925-9yfygtcm7fwsuq2p-1 mysql-test/r/subselect_sj.result subselect_sj.result-20100117143926-nrop4ku355g3kv8b-1 mysql-test/r/subselect_sj_jcl6.result subselect_sj_jcl6.re-20100117143928-7vzk51yaf29cdavp-1 mysql-test/t/ps.test sp1f-ps.test-20040405154119-4zqf6po44yypvz5foa2osprg5kb5ok63 mysql-test/t/subselect.test sp1f-subselect.test-20020512204640-lyqrayx6uwsn7zih6y7kerkenuitzbvr mysql-test/t/subselect3.test sp1f-subselect3.test-20061031174245-pcxt5ljylerxhx2jkfhrbqfv5vqcazlz sql/item_cmpfunc.h sp1f-item_cmpfunc.h-19700101030959-pcvbjplo4e4ng7ibynfhcd6pjyem57gr sql/item_subselect.cc sp1f-item_subselect.cc-20020512204640-qep43aqhsfrwkqmrobni6czc3fqj36oo sql/item_subselect.h sp1f-item_subselect.h-20020512204640-qdg77wil56cxyhtc2bjjdrppxq3wqgh3 sql/mysql_priv.h sp1f-mysql_priv.h-19700101030959-4fl65tqpop5zfgxaxkqotu2fa2ree5ci sql/mysqld.cc sp1f-mysqld.cc-19700101030959-zpswdvekpvixxzxf7gdtofzel7nywtfj sql/opt_subselect.cc opt_subselect.cc-20100215190428-nekkl8wisp0k6nlk-1 sql/set_var.cc sp1f-set_var.cc-20020723153119-nwbpg2pwpz55pfw7yfzaxt7hsszzy7y3 sql/sql_class.cc sp1f-sql_class.cc-19700101030959-rpotnweaff2pikkozh3butrf7mv3oero sql/sql_class.h sp1f-sql_class.h-19700101030959-jnqnbrjyqsvgncsibnumsmg3lyi7pa5s sql/sql_select.cc sp1f-sql_select.cc-19700101030959-egb7whpkh76zzvikycs5nsnuviu4fdlb === modified file 'mysql-test/include/mix1.inc' --- a/mysql-test/include/mix1.inc 2009-09-15 06:08:54 +0000 +++ b/mysql-test/include/mix1.inc 2010-03-11 21:43:31 +0000 @@ -1177,8 +1177,11 @@ create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; --echo End of 5.0 tests === modified file 'mysql-test/r/index_merge_myisam.result' --- a/mysql-test/r/index_merge_myisam.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/index_merge_myisam.result 2010-03-11 21:43:31 +0000 @@ -1419,19 +1419,19 @@ # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge=off,index_merge_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge_union=on'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,index_merge_sort_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=4; ERROR 42000: Variable 'optimizer_switch' can't be set to the value of '4' set optimizer_switch=NULL; @@ -1458,21 +1458,21 @@ set optimizer_switch='index_merge=off,index_merge_union=off,default'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set @@global.optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on # # Check index_merge's @@optimizer_switch flags # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, c int, filler char(100), @@ -1582,5 +1582,5 @@ set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on drop table t0, t1; === modified file 'mysql-test/r/innodb_mysql.result' --- a/mysql-test/r/innodb_mysql.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mysql.result 2010-03-11 21:43:31 +0000 @@ -1425,12 +1425,15 @@ # create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables 2 DEPENDENT SUBQUERY t1 system NULL NULL NULL NULL 0 const row not found 2 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 1 +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; End of 5.0 tests CREATE TABLE `t2` ( === modified file 'mysql-test/r/myisam_mrr.result' --- a/mysql-test/r/myisam_mrr.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/myisam_mrr.result 2010-03-11 21:43:31 +0000 @@ -394,7 +394,7 @@ # - engine_condition_pushdown does not affect ICP select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, key(a)); === modified file 'mysql-test/r/ps.result' --- a/mysql-test/r/ps.result 2009-05-27 15:19:44 +0000 +++ b/mysql-test/r/ps.result 2010-03-11 21:43:31 +0000 @@ -149,6 +149,8 @@ c32 set('monday', 'tuesday', 'wednesday') ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -177,6 +179,7 @@ 2 DEPENDENT SUBQUERY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; set @arg00=1; prepare stmt1 from ' create table t1 (m int) as select 1 as m ' ; execute stmt1 ; === modified file 'mysql-test/r/subselect.result' --- a/mysql-test/r/subselect.result 2010-02-17 21:59:41 +0000 +++ b/mysql-test/r/subselect.result 2010-03-11 21:43:31 +0000 @@ -1,4 +1,6 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4803,4 +4805,5 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. === modified file 'mysql-test/r/subselect3.result' --- a/mysql-test/r/subselect3.result 2010-02-17 10:05:27 +0000 +++ b/mysql-test/r/subselect3.result 2010-03-11 21:43:31 +0000 @@ -63,12 +63,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -692,6 +695,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -759,6 +764,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -960,7 +966,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -977,7 +983,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect3_jcl6.result' --- a/mysql-test/r/subselect3_jcl6.result 2010-02-17 10:47:55 +0000 +++ b/mysql-test/r/subselect3_jcl6.result 2010-03-11 21:43:31 +0000 @@ -67,12 +67,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -696,6 +699,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -763,6 +768,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -964,7 +970,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -981,7 +987,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect_no_mat.result' --- a/mysql-test/r/subselect_no_mat.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_mat.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_opts.result' --- a/mysql-test/r/subselect_no_opts.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_opts.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off,semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_semijoin.result' --- a/mysql-test/r/subselect_no_semijoin.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_semijoin.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-15 19:52:58 +0000 @@ -202,39 +202,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 19:52:58 +0000 @@ -206,39 +206,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/t/ps.test' --- a/mysql-test/t/ps.test 2009-05-27 15:19:44 +0000 +++ b/mysql-test/t/ps.test 2010-03-11 21:43:31 +0000 @@ -163,6 +163,9 @@ ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -171,6 +174,8 @@ deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # parameters from variables (for field creation) # === modified file 'mysql-test/t/subselect.test' --- a/mysql-test/t/subselect.test 2010-01-17 20:52:20 +0000 +++ b/mysql-test/t/subselect.test 2010-03-11 21:43:31 +0000 @@ -11,6 +11,9 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; --enable_warnings +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + select (select 2); explain extended select (select 2); SELECT (SELECT 1) UNION SELECT (SELECT 2); @@ -4061,4 +4064,6 @@ (SELECT LAST_INSERT_ID() FROM t1 ORDER BY MIN(a) ASC LIMIT 1); DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; + --echo End of 5.1 tests. === modified file 'mysql-test/t/subselect3.test' --- a/mysql-test/t/subselect3.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect3.test 2010-03-11 21:43:31 +0000 @@ -59,9 +59,13 @@ show status like 'Handler_read_rnd_next'; select ' ^ This must show 11' Z; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + # This must show trigcond: explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; # @@ -529,6 +533,9 @@ DROP TABLE t1, t2; +# The next three test cases must be executed with the IN=>EXISTS strategy +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; # # Bug #27870: crash of an equijoin query with WHERE condition containing @@ -588,6 +595,8 @@ DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #34763: item_subselect.cc:1235:Item_in_subselect::row_value_transformer: # Assertion failed, unexpected error message: === modified file 'sql/item_cmpfunc.h' --- a/sql/item_cmpfunc.h 2010-03-13 20:04:52 +0000 +++ b/sql/item_cmpfunc.h 2010-03-15 19:52:58 +0000 @@ -350,6 +350,7 @@ CHARSET_INFO *compare_collation() { return cmp.cmp_collation.collation; } uint decimal_precision() const { return 1; } void top_level_item() { abort_on_null= TRUE; } + Arg_comparator *get_comparator() { return &cmp; } friend class Arg_comparator; }; === modified file 'sql/item_subselect.cc' --- a/sql/item_subselect.cc 2010-02-21 06:32:23 +0000 +++ b/sql/item_subselect.cc 2010-03-09 10:14:06 +0000 @@ -138,6 +138,7 @@ left_expr_cache= NULL; } first_execution= TRUE; + is_constant= FALSE; Item_subselect::cleanup(); DBUG_VOID_RETURN; } @@ -449,8 +450,10 @@ int res; if (thd->is_error()) - /* Do not execute subselect in case of a fatal error */ + { + /* Do not execute subselect in case of a fatal error */ return 1; + } /* Simulate a failure in sub-query execution. Used to test e.g. out of memory or query being killed conditions. @@ -475,9 +478,6 @@ bool Item_in_subselect::exec() { DBUG_ENTER("Item_in_subselect::exec"); - DBUG_ASSERT(exec_method != MATERIALIZATION || - (exec_method == MATERIALIZATION && - engine->engine_type() == subselect_engine::HASH_SJ_ENGINE)); /* Initialize the cache of the left predicate operand. This has to be done as late as now, because Cached_item directly contains a resolved field (not @@ -493,14 +493,14 @@ if (!left_expr_cache && exec_method == MATERIALIZATION) init_left_expr_cache(); - /* If the new left operand is already in the cache, reuse the old result. */ - if (left_expr_cache && test_if_item_cache_changed(*left_expr_cache) < 0) - { - /* Always compute IN for the first row as the cache is not valid for it. */ - if (!first_execution) - DBUG_RETURN(FALSE); - first_execution= FALSE; - } + /* + If the new left operand is already in the cache, reuse the old result. + Use the cached result only if this is not the first execution of IN + because the cache is not valid for the first execution. + */ + if (!first_execution && left_expr_cache && + test_if_item_cache_changed(*left_expr_cache) < 0) + DBUG_RETURN(FALSE); /* The exec() method below updates item::value, and item::null_value, thus if @@ -910,8 +910,8 @@ Item_in_subselect::Item_in_subselect(Item * left_exp, st_select_lex *select_lex): Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), - upper_item(0) + is_constant(FALSE), optimizer(0), pushed_cond_guards(NULL), + exec_method(NOT_TRANSFORMED), upper_item(0) { DBUG_ENTER("Item_in_subselect::Item_in_subselect"); left_expr= left_exp; @@ -1105,6 +1105,8 @@ { DBUG_ASSERT(fixed == 1); null_value= 0; + if (is_constant) + return value; if (exec()) { reset(); @@ -1571,9 +1573,9 @@ DBUG_ENTER("Item_in_subselect::row_value_transformer"); // psergey: duplicated_subselect_card_check - if (select_lex->item_list.elements != left_expr->cols()) + if (select_lex->item_list.elements != cols_num) { - my_error(ER_OPERAND_COLUMNS, MYF(0), left_expr->cols()); + my_error(ER_OPERAND_COLUMNS, MYF(0), cols_num); DBUG_RETURN(RES_ERROR); } @@ -1980,17 +1982,69 @@ bool Item_in_subselect::fix_fields(THD *thd_arg, Item **ref) { - bool result = 0; + uint outer_cols_num; + List<Item> *inner_cols; if (exec_method == SEMI_JOIN) return !( (*ref)= new Item_int(1)); - if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed) - result = left_expr->fix_fields(thd_arg, &left_expr); - - return result || Item_subselect::fix_fields(thd_arg, ref); + /* + Check if the outer and inner IN operands match in those cases when we + will not perform IN=>EXISTS transformation. Currently this is when we + use subquery materialization. + + The condition below is true when this method was called recursively from + inside JOIN::prepare for the JOIN object created by the call chain + Item_subselect::fix_fields -> subselect_single_select_engine::prepare, + which creates a JOIN object for the subquery and calls JOIN::prepare for + the JOIN of the subquery. + Notice that in some cases, this doesn't happen, and the check_cols() + test for each Item happens later in + Item_in_subselect::row_value_in_to_exists_transformer. + The reason for this mess is that our JOIN::prepare phase works top-down + instead of bottom-up, so we first do name resoluton and semantic checks + for the outer selects, then for the inner. + */ + if (engine && + engine->engine_type() == subselect_engine::SINGLE_SELECT_ENGINE && + ((subselect_single_select_engine*)engine)->join) + { + outer_cols_num= left_expr->cols(); + + if (unit->is_union()) + inner_cols= &(unit->types); + else + inner_cols= &(unit->first_select()->item_list); + if (outer_cols_num != inner_cols->elements) + { + my_error(ER_OPERAND_COLUMNS, MYF(0), outer_cols_num); + return TRUE; + } + if (outer_cols_num > 1) + { + List_iterator<Item> inner_col_it(*inner_cols); + Item *inner_col; + for (uint i= 0; i < outer_cols_num; i++) + { + inner_col= inner_col_it++; + if (inner_col->check_cols(left_expr->element_index(i)->cols())) + return TRUE; + } + } + } + + if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed && + left_expr->fix_fields(thd_arg, &left_expr)) + return TRUE; + if (Item_subselect::fix_fields(thd_arg, ref)) + return TRUE; + + fixed= TRUE; + + return FALSE; } + void Item_in_subselect::fix_after_pullout(st_select_lex *new_parent, Item **ref) { left_expr->fix_after_pullout(new_parent, &left_expr); @@ -2267,10 +2321,9 @@ void subselect_uniquesubquery_engine::cleanup() { DBUG_ENTER("subselect_uniquesubquery_engine::cleanup"); - /* - subselect_uniquesubquery_engine have not 'result' assigbed, so we do not - cleanup() it - */ + /* Tell handler we don't need the index anymore */ + if (tab->table->file->inited) + tab->table->file->ha_index_end(); DBUG_VOID_RETURN; } @@ -2291,7 +2344,7 @@ Create and prepare the JOIN object that represents the query execution plan for the subquery. - @detail + @details This method is called from Item_subselect::fix_fields. For prepared statements it is called both during the PREPARE and EXECUTE phases in the following ways: @@ -2593,14 +2646,23 @@ for (;;) { error=table->file->ha_rnd_next(table->record[0]); - if (error && error != HA_ERR_END_OF_FILE) - { - error= report_error(table, error); - break; + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(table, error); + break; + } } - /* No more rows */ - if (table->status) - break; if (!cond || cond->val_int()) { @@ -2711,6 +2773,56 @@ /* + @retval 1 A NULL was found in the outer reference, index lookup is + not applicable, the outer ref is unsusable as a lookup key, + use some other method to find a match. + @retval 0 The outer ref was copied into an index lookup key. + @retval -1 The outer ref cannot possibly match any row, IN is FALSE. +*/ +/* TIMOUR: this method is a variant of copy_ref_key(), needs refactoring. */ + +int subselect_uniquesubquery_engine::copy_ref_key_simple() +{ + for (store_key **copy= tab->ref.key_copy ; *copy ; copy++) + { + enum store_key::store_key_result store_res; + store_res= (*copy)->copy(); + tab->ref.key_err= store_res; + + /* + When there is a NULL part in the key we don't need to make index + lookup for such key thus we don't need to copy whole key. + If we later should do a sequential scan return OK. Fail otherwise. + + See also the comment for the subselect_uniquesubquery_engine::exec() + function. + */ + null_keypart= (*copy)->null_key; + if (null_keypart) + return 1; + + /* + Check if the error is equal to STORE_KEY_FATAL. This is not expressed + using the store_key::store_key_result enum because ref.key_err is a + boolean and we want to detect both TRUE and STORE_KEY_FATAL from the + space of the union of the values of [TRUE, FALSE] and + store_key::store_key_result. + TODO: fix the variable an return types. + */ + if (store_res == store_key::STORE_KEY_FATAL) + { + /* + Error converting the left IN operand to the column type of the right + IN operand. + */ + return -1; + } + } + return 0; +} + + +/* Execute subselect SYNOPSIS @@ -2750,7 +2862,13 @@ /* TODO: change to use of 'full_scan' here? */ if (copy_ref_key()) + { + /* + TIMOUR: copy_ref_key() == 1 means NULL result, not error, why return 1? + Check who reiles on this result. + */ DBUG_RETURN(1); + } if (table->status) { /* @@ -2791,6 +2909,46 @@ } +/* + TIMOUR: write comment +*/ + +int subselect_uniquesubquery_engine::index_lookup() +{ + DBUG_ENTER("subselect_uniquesubquery_engine::index_lookup"); + int error; + TABLE *table= tab->table; + + if (!table->file->inited) + table->file->ha_index_init(tab->ref.key, 0); + error= table->file->ha_index_read_map(table->record[0], + tab->ref.key_buff, + make_prev_keypart_map(tab-> + ref.key_parts), + HA_READ_KEY_EXACT); + DBUG_PRINT("info", ("lookup result: %i", error)); + + if (error && error != HA_ERR_KEY_NOT_FOUND && error != HA_ERR_END_OF_FILE) + { + /* + TIMOUR: I don't understand at all when do we need to call report_error. + In most places where we access an index, we don't do this. Why here? + */ + error= report_error(table, error); + DBUG_RETURN(error); + } + + table->null_row= 0; + if (!error && (!cond || cond->val_int())) + ((Item_in_subselect *) item)->value= 1; + else + ((Item_in_subselect *) item)->value= 0; + + DBUG_RETURN(0); +} + + + subselect_uniquesubquery_engine::~subselect_uniquesubquery_engine() { /* Tell handler we don't need the index anymore */ @@ -3225,6 +3383,7 @@ bool subselect_uniquesubquery_engine::no_tables() { /* returning value is correct, but this method should never be called */ + DBUG_ASSERT(FALSE); return 0; } @@ -3235,16 +3394,259 @@ /** + Check if an IN predicate should be executed via partial matching using + only schema information. + + @details + This test essentially has three results: + - partial matching is applicable, but cannot be executed due to a + limitation in the total number of indexes, as a result we can't + use subquery materialization at all. + - partial matching is either applicable or not, and this can be + determined by looking at 'this->max_keys'. + If max_keys > 1, then we need partial matching because there are + more indexes than just the one we use during materialization to + remove duplicates. + + @note + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_schema() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + + if (item_in->is_top_level_item()) + return COMPLETE_MATCH; + else + { + List_iterator<Item> inner_col_it(*item_in->unit->get_unit_column_types()); + Item *outer_col, *inner_col; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + outer_col= item_in->left_expr->element_index(i); + inner_col= inner_col_it++; + + if (!inner_col->maybe_null && !outer_col->maybe_null) + bitmap_set_bit(&non_null_key_parts, i); + else + { + bitmap_set_bit(&partial_match_key_parts, i); + ++count_partial_match_columns; + } + } + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (count_partial_match_columns) + return PARTIAL_MATCH; + return COMPLETE_MATCH; +} + + +/** + Test whether an IN predicate must be computed via partial matching + based on the NULL statistics for each column of a materialized subquery. + + @details The procedure analyzes column NULL statistics, updates the + matching type of columns that cannot be NULL or that contain only NULLs. + Based on this, the procedure determines the final execution strategy for + the [NOT] IN predicate. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_data() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + Item *outer_col; + + /* + If we already determined that a complete match is enough based on schema + information, nothing can be better. + */ + if (strategy == COMPLETE_MATCH) + return COMPLETE_MATCH; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + if (!bitmap_is_set(&partial_match_key_parts, i)) + continue; + outer_col= item_in->left_expr->element_index(i); + /* + If column 'i' doesn't contain NULLs, and the corresponding outer reference + cannot have a NULL value, then 'i' is a non-nullable column. + */ + if (result_sink->get_null_count_of_col(i) == 0 && !outer_col->maybe_null) + { + bitmap_clear_bit(&partial_match_key_parts, i); + bitmap_set_bit(&non_null_key_parts, i); + --count_partial_match_columns; + } + if (result_sink->get_null_count_of_col(i) == + tmp_table->file->stats.records) + ++count_null_only_columns; + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (!count_partial_match_columns) + return COMPLETE_MATCH; + return PARTIAL_MATCH; +} + + +void +subselect_hash_sj_engine::choose_partial_match_strategy( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t pm_buff_size; + + DBUG_ASSERT(strategy == PARTIAL_MATCH); + /* + Choose according to global optimizer switch. If only one of the switches is + 'ON', then the remaining strategy is the only possible one. The only cases + when this will be overriden is when the total size of all buffers for the + merge strategy is bigger than the 'rowid_merge_buff_size' system variable, + or if there isn't enough physical memory to allocate the buffers. + */ + if (!optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_SCAN; + else if + ( optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + !optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_MERGE; + + /* + If both switches are ON, or both are OFF, we interpret that as "let the + optimizer decide". Perform a cost based choice between the two partial + matching strategies. + */ + /* + TIMOUR: the above interpretation of the switch values could be changed to: + - if both are ON - let the optimizer decide, + - if both are OFF - do not use partial matching, therefore do not use + materialization in non-top-level predicates. + The problem with this is that we know for sure if we need partial matching + only after the subquery is materialized, and this is too late to revert to + the IN=>EXISTS strategy. + */ + if (strategy == PARTIAL_MATCH) + { + /* + TIMOUR: Currently we use a super simplistic measure. This will be + addressed in a separate task. + */ + if (tmp_table->file->stats.records < 100) + strategy= PARTIAL_MATCH_SCAN; + else + strategy= PARTIAL_MATCH_MERGE; + } + + /* Check if there is enough memory for the rowid merge strategy. */ + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_buff_size= rowid_merge_buff_size(has_non_null_key, + has_covering_null_row, + partial_match_key_parts); + if (pm_buff_size > thd->variables.rowid_merge_buff_size) + strategy= PARTIAL_MATCH_SCAN; + } +} + + +/* + Compute the memory size of all buffers proportional to the number of rows + in tmp_table. + + @details + If the result is bigger than thd->variables.rowid_merge_buff_size, partial + matching via merging is not applicable. +*/ + +size_t subselect_hash_sj_engine::rowid_merge_buff_size( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t buff_size; /* Total size of all buffers used by partial matching. */ + ha_rows row_count= tmp_table->file->stats.records; + uint rowid_length= tmp_table->file->ref_length; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + /* Size of the subselect_rowid_merge_engine::row_num_to_rowid buffer. */ + buff_size= row_count * rowid_length * sizeof(uchar); + + if (has_non_null_key) + { + /* Add the size of Ordered_key::key_buff of the only non-NULL key. */ + buff_size+= row_count * sizeof(rownum_t); + } + + if (!has_covering_null_row) + { + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i) || + result_sink->get_null_count_of_col(i) == row_count) + continue; /* In these cases we wouldn't construct Ordered keys. */ + + /* Add the size of Ordered_key::key_buff */ + buff_size+= (row_count - result_sink->get_null_count_of_col(i)) * + sizeof(rownum_t); + /* Add the size of Ordered_key::null_key */ + buff_size+= bitmap_buffer_size(result_sink->get_max_null_of_col(i)); + } + } + + return buff_size; +} + + +/* + Initialize a MY_BITMAP with a buffer allocated on the current + memory root. + TIMOUR: move to bitmap C file? +*/ + +static my_bool +bitmap_init_memroot(MY_BITMAP *map, uint n_bits, MEM_ROOT *mem_root) +{ + my_bitmap_map *bitmap_buf; + + if (!(bitmap_buf= (my_bitmap_map*) alloc_root(mem_root, + bitmap_buffer_size(n_bits))) || + bitmap_init(map, bitmap_buf, n_bits, FALSE)) + return TRUE; + bitmap_clear_all(map); + return FALSE; +} + + +/** Create all structures needed for IN execution that can live between PS reexecution. - @detail + @param tmp_columns the items that produce the data for the temp table + + @details - Create a temporary table to store the result of the IN subquery. The temporary table has one hash index on all its columns. - Create a new result sink that sends the result stream of the subquery to the temporary table, - - Create and initialize a new JOIN_TAB, and TABLE_REF objects to perform - lookups into the indexed temporary table. @notice: Currently Item_subselect::init() already chooses and creates at parse @@ -3256,71 +3658,178 @@ bool subselect_hash_sj_engine::init_permanent(List<Item> *tmp_columns) { - /* The result sink where we will materialize the subquery result. */ - select_union *tmp_result_sink; - /* The table into which the subquery is materialized. */ - TABLE *tmp_table; - KEY *tmp_key; /* The only index on the temporary table. */ - uint tmp_key_parts; /* Number of keyparts in tmp_key. */ - Item_in_subselect *item_in= (Item_in_subselect *) item; + /* Options to create_tmp_table. */ + ulonglong tmp_create_options= thd->options | TMP_TABLE_ALL_COLUMNS; + /* | TMP_TABLE_FORCE_MYISAM; TIMOUR: force MYISAM */ DBUG_ENTER("subselect_hash_sj_engine::init_permanent"); - /* 1. Create/initialize materialization related objects. */ + if (bitmap_init_memroot(&non_null_key_parts, tmp_columns->elements, + thd->mem_root) || + bitmap_init_memroot(&partial_match_key_parts, tmp_columns->elements, + thd->mem_root)) + DBUG_RETURN(TRUE); /* Create and initialize a select result interceptor that stores the result stream in a temporary table. The temporary table itself is managed (created/filled/etc) internally by the interceptor. */ - if (!(tmp_result_sink= new select_union)) - DBUG_RETURN(TRUE); - if (tmp_result_sink->create_result_table( - thd, tmp_columns, TRUE, - thd->options | TMP_TABLE_ALL_COLUMNS, +/* + TIMOUR: + Select a more efficient result sink when we know there is no need to collect + data statistics. + + if (strategy == COMPLETE_MATCH) + { + if (!(result= new select_union)) + DBUG_RETURN(TRUE); + } + else if (strategy == PARTIAL_MATCH) + { + if (!(result= new select_materialize_with_stats)) + DBUG_RETURN(TRUE); + } +*/ + if (!(result= new select_materialize_with_stats)) + DBUG_RETURN(TRUE); + + if (((select_union*) result)->create_result_table( + thd, tmp_columns, TRUE, tmp_create_options, "materialized subselect", TRUE)) DBUG_RETURN(TRUE); - tmp_table= tmp_result_sink->table; - tmp_key= tmp_table->key_info; - tmp_key_parts= tmp_key->key_parts; + tmp_table= ((select_union*) result)->table; /* - If the subquery has blobs, or the total key lenght is bigger than some - length, then the created index cannot be used for lookups and we - can't use hash semi join. If this is the case, delete the temporary - table since it will not be used, and tell the caller we failed to - initialize the engine. + If the subquery has blobs, or the total key lenght is bigger than + some length, or the total number of key parts is more than the + allowed maximum (currently MAX_REF_PARTS == 16), then the created + index cannot be used for lookups and we can't use hash semi + join. If this is the case, delete the temporary table since it + will not be used, and tell the caller we failed to initialize the + engine. */ if (tmp_table->s->keys == 0) { -#ifndef DBUG_OFF - handlerton *tmp_table_hton= tmp_table->s->db_type(); -#ifdef USE_MARIA_FOR_TMP_TABLES - DBUG_ASSERT(tmp_table_hton == maria_hton); -#else - DBUG_ASSERT(tmp_table_hton == myisam_hton); -#endif -#endif DBUG_ASSERT( tmp_table->s->uniques || tmp_table->key_info->key_length >= tmp_table->file->max_key_length() || tmp_table->key_info->key_parts > tmp_table->file->max_key_parts()); free_tmp_table(thd, tmp_table); + tmp_table= NULL; delete result; result= NULL; DBUG_RETURN(TRUE); } - result= tmp_result_sink; /* Make sure there is only one index on the temp table, and it doesn't have the extra key part created when s->uniques > 0. */ - DBUG_ASSERT(tmp_table->s->keys == 1 && tmp_columns->elements == tmp_key_parts); - - - /* 2. Create/initialize execution related objects. */ + DBUG_ASSERT(tmp_table->s->keys == 1 && + ((Item_in_subselect *) item)->left_expr->cols() == + tmp_table->key_info->key_parts); + + if (make_semi_join_conds() || + /* A unique_engine is used both for complete and partial matching. */ + !(lookup_engine= make_unique_engine())) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} + + +/* + Create an artificial condition to post-filter those rows matched by index + lookups that cannot be distinguished by the index lookup procedure. + + @notes + The need for post-filtering may occur e.g. because of + truncation. Prepared statements execution requires that fix_fields is + called for every execution. In order to call fix_fields we need to + create a Name_resolution_context and a corresponding TABLE_LIST for + the temporary table for the subquery, so that all column references + to the materialized subquery table can be resolved correctly. + + @returns + @retval TRUE memory allocation error occurred + @retval FALSE the conditions were created and resolved (fixed) +*/ + +bool subselect_hash_sj_engine::make_semi_join_conds() +{ + /* + Table reference for tmp_table that is used to resolve column references + (Item_fields) to columns in tmp_table. + */ + TABLE_LIST *tmp_table_ref; + /* Name resolution context for all tmp_table columns created below. */ + Name_resolution_context *context; + Item_in_subselect *item_in= (Item_in_subselect *) item; + + DBUG_ENTER("subselect_hash_sj_engine::make_semi_join_conds"); + DBUG_ASSERT(semi_join_conds == NULL); + + if (!(semi_join_conds= new Item_cond_and)) + DBUG_RETURN(TRUE); + + if (!(tmp_table_ref= (TABLE_LIST*) thd->alloc(sizeof(TABLE_LIST)))) + DBUG_RETURN(TRUE); + + tmp_table_ref->init_one_table("", "materialized subselect", TL_READ); + tmp_table_ref->table= tmp_table; + + context= new Name_resolution_context; + context->init(); + context->first_name_resolution_table= + context->last_name_resolution_table= tmp_table_ref; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + Item_func_eq *eq_cond; /* New equi-join condition for the current column. */ + /* Item for the corresponding field from the materialized temp table. */ + Item_field *right_col_item; + + if (!(right_col_item= new Item_field(thd, context, tmp_table->field[i])) || + !(eq_cond= new Item_func_eq(item_in->left_expr->element_index(i), + right_col_item)) || + (((Item_cond_and*)semi_join_conds)->add(eq_cond))) + { + delete semi_join_conds; + semi_join_conds= NULL; + DBUG_RETURN(TRUE); + } + } + if (semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} + + +/** + Create a new uniquesubquery engine for the execution of an IN predicate. + + @details + Create and initialize a new JOIN_TAB, and Table_ref objects to perform + lookups into the indexed temporary table. + + @retval A new subselect_hash_sj_engine object + @retval NULL if a memory allocation error occurs +*/ + +subselect_uniquesubquery_engine* +subselect_hash_sj_engine::make_unique_engine() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + /* The only index on the temporary table. */ + KEY *tmp_key= tmp_table->key_info; + /* Number of keyparts in tmp_key. */ + uint tmp_key_parts= tmp_key->key_parts; + JOIN_TAB *tab; + + DBUG_ENTER("subselect_hash_sj_engine::make_unique_engine"); /* Create and initialize the JOIN_TAB that represents an index lookup @@ -3328,9 +3837,9 @@ - this JOIN_TAB has no corresponding JOIN (and doesn't need one), and - here we initialize only those members that are used by subselect_uniquesubquery_engine, so these objects are incomplete. - */ + */ if (!(tab= (JOIN_TAB*) thd->alloc(sizeof(JOIN_TAB)))) - DBUG_RETURN(TRUE); + DBUG_RETURN(NULL); tab->table= tmp_table; tab->ref.key= 0; /* The only temp table index. */ tab->ref.key_length= tmp_key->key_length; @@ -3341,60 +3850,18 @@ (tmp_key_parts + 1)))) || !(tab->ref.items= (Item**) thd->alloc(sizeof(Item*) * tmp_key_parts))) - DBUG_RETURN(TRUE); + DBUG_RETURN(NULL); KEY_PART_INFO *cur_key_part= tmp_key->key_part; store_key **ref_key= tab->ref.key_copy; uchar *cur_ref_buff= tab->ref.key_buff; - - /* - Create an artificial condition to post-filter those rows matched by index - lookups that cannot be distinguished by the index lookup procedure, e.g. - because of truncation. Prepared statements execution requires that - fix_fields is called for every execution. In order to call fix_fields we - need to create a Name_resolution_context and a corresponding TABLE_LIST - for the temporary table for the subquery, so that all column references - to the materialized subquery table can be resolved correctly. - */ - DBUG_ASSERT(cond == NULL); - if (!(cond= new Item_cond_and)) - DBUG_RETURN(TRUE); - /* - Table reference for tmp_table that is used to resolve column references - (Item_fields) to columns in tmp_table. - */ - TABLE_LIST *tmp_table_ref; - if (!(tmp_table_ref= (TABLE_LIST*) thd->alloc(sizeof(TABLE_LIST)))) - DBUG_RETURN(TRUE); - - tmp_table_ref->init_one_table("", "materialized subselect", TL_READ); - tmp_table_ref->table= tmp_table; - - /* Name resolution context for all tmp_table columns created below. */ - Name_resolution_context *context= new Name_resolution_context; - context->init(); - context->first_name_resolution_table= - context->last_name_resolution_table= tmp_table_ref; for (uint i= 0; i < tmp_key_parts; i++, cur_key_part++, ref_key++) { - Item_func_eq *eq_cond; /* New equi-join condition for the current column. */ - /* Item for the corresponding field from the materialized temp table. */ - Item_field *right_col_item; + tab->ref.items[i]= item_in->left_expr->element_index(i); int null_count= test(cur_key_part->field->real_maybe_null()); - tab->ref.items[i]= item_in->left_expr->element_index(i); - - if (!(right_col_item= new Item_field(thd, context, cur_key_part->field)) || - !(eq_cond= new Item_func_eq(tab->ref.items[i], right_col_item)) || - ((Item_cond_and*)cond)->add(eq_cond)) - { - delete cond; - cond= NULL; - DBUG_RETURN(TRUE); - } - *ref_key= new store_key_item(thd, cur_key_part->field, - /* TODO: + /* TIMOUR: the NULL byte is taken into account in cur_key_part->store_length, so instead of cur_ref_buff + test(maybe_null), we could @@ -3409,10 +3876,8 @@ tab->ref.key_err= 1; tab->ref.key_parts= tmp_key_parts; - if (cond->fix_fields(thd, &cond)) - DBUG_RETURN(TRUE); - - DBUG_RETURN(FALSE); + DBUG_RETURN(new subselect_uniquesubquery_engine(thd, tab, item, + semi_join_conds)); } @@ -3435,7 +3900,8 @@ Repeat name resolution for 'cond' since cond is not part of any clause of the query, and it is not 'fixed' during JOIN::prepare. */ - if (cond && !cond->fixed && cond->fix_fields(thd, &cond)) + if (semi_join_conds && !semi_join_conds->fixed && + semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) return TRUE; /* Let our engine reuse this query plan for materialization. */ materialize_join= materialize_engine->join; @@ -3446,32 +3912,53 @@ subselect_hash_sj_engine::~subselect_hash_sj_engine() { + delete lookup_engine; delete result; - if (tab) - free_tmp_table(thd, tab->table); + if (tmp_table) + free_tmp_table(thd, tmp_table); } /** Cleanup performed after each PS execution. - @detail + @details Called in the end of JOIN::prepare for PS from Item_subselect::cleanup. */ void subselect_hash_sj_engine::cleanup() { + enum_engine_type lookup_engine_type= lookup_engine->engine_type(); is_materialized= FALSE; + bitmap_clear_all(&non_null_key_parts); + bitmap_clear_all(&partial_match_key_parts); + count_partial_match_columns= 0; + count_null_only_columns= 0; + strategy= UNDEFINED; + materialize_engine->cleanup(); + if (lookup_engine_type == TABLE_SCAN_ENGINE || + lookup_engine_type == ROWID_MERGE_ENGINE) + { + subselect_engine *inner_lookup_engine; + inner_lookup_engine= + ((subselect_partial_match_engine*) lookup_engine)->lookup_engine; + /* + Partial match engines are recreated for each PS execution inside + subselect_hash_sj_engine::exec(). + */ + delete lookup_engine; + lookup_engine= inner_lookup_engine; + } + DBUG_ASSERT(lookup_engine->engine_type() == UNIQUESUBQUERY_ENGINE); + lookup_engine->cleanup(); result->cleanup(); /* Resets the temp table as well. */ - materialize_engine->cleanup(); - subselect_uniquesubquery_engine::cleanup(); } /** Execute a subquery IN predicate via materialization. - @detail + @details If needed materialize the subquery into a temporary table, then copmpute the predicate via a lookup into this table. @@ -3482,6 +3969,9 @@ int subselect_hash_sj_engine::exec() { Item_in_subselect *item_in= (Item_in_subselect *) item; + SELECT_LEX *save_select= thd->lex->current_select; + subselect_partial_match_engine *pm_engine= NULL; + int res= 0; DBUG_ENTER("subselect_hash_sj_engine::exec"); @@ -3489,56 +3979,126 @@ Optimize and materialize the subquery during the first execution of the subquery predicate. */ - if (!is_materialized) - { - int res= 0; - SELECT_LEX *save_select= thd->lex->current_select; - thd->lex->current_select= materialize_engine->select_lex; - if ((res= materialize_join->optimize())) - goto err; /* purecov: inspected */ - materialize_join->exec(); - if ((res= test(materialize_join->error || thd->is_fatal_error))) - goto err; - - /* - TODO: - - Unlock all subquery tables as we don't need them. To implement this - we need to add new functionality to JOIN::join_free that can unlock - all tables in a subquery (and all its subqueries). - - The temp table used for grouping in the subquery can be freed - immediately after materialization (yet it's done together with - unlocking). - */ - is_materialized= TRUE; - /* - If the subquery returned no rows, the temporary table is empty, so we know - directly that the result of IN is FALSE. We first update the table - statistics, then we test if the temporary table for the query result is - empty. - */ - tab->table->file->info(HA_STATUS_VARIABLE); - if (!tab->table->file->stats.records) - { - empty_result_set= TRUE; - item_in->value= FALSE; - /* TODO: check we need this: item_in->null_value= FALSE; */ - DBUG_RETURN(FALSE); - } - /* Set tmp_param only if its usable, i.e. tmp_param->copy_field != NULL. */ - tmp_param= &(item_in->unit->outer_select()->join->tmp_table_param); - if (tmp_param && !tmp_param->copy_field) - tmp_param= NULL; + thd->lex->current_select= materialize_engine->select_lex; + if ((res= materialize_join->optimize())) + goto err; /* purecov: inspected */ + DBUG_ASSERT(!is_materialized); /* We should materialize only once. */ + materialize_join->exec(); + if ((res= test(materialize_join->error || thd->is_fatal_error))) + goto err; + + /* + TODO: + - Unlock all subquery tables as we don't need them. To implement this + we need to add new functionality to JOIN::join_free that can unlock + all tables in a subquery (and all its subqueries). + - The temp table used for grouping in the subquery can be freed + immediately after materialization (yet it's done together with + unlocking). + */ + is_materialized= TRUE; + /* + If the subquery returned no rows, the temporary table is empty, so we know + directly that the result of IN is FALSE. We first update the table + statistics, then we test if the temporary table for the query result is + empty. + */ + tmp_table->file->info(HA_STATUS_VARIABLE); + if (!tmp_table->file->stats.records) + { + item_in->value= FALSE; + /* The value of IN will not change during this execution. */ + item_in->is_constant= TRUE; + item_in->set_first_execution(); + /* TIMOUR: check if we need this: item_in->null_value= FALSE; */ + DBUG_RETURN(FALSE); + } + + /* + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. + */ + strategy= get_strategy_using_schema(); + /* This call may discover that we don't need partial matching at all. */ + strategy= get_strategy_using_data(); + if (strategy == PARTIAL_MATCH) + { + uint count_pm_keys; /* Total number of keys needed for partial matching. */ + MY_BITMAP *nn_key_parts; /* The key parts of the only non-NULL index. */ + uint covering_null_row_width; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + nn_key_parts= (count_partial_match_columns < tmp_table->s->fields) ? + &non_null_key_parts : NULL; + + if (result_sink->get_max_nulls_in_row() == + tmp_table->s->fields - + (nn_key_parts ? bitmap_bits_set(nn_key_parts) : 0)) + covering_null_row_width= result_sink->get_max_nulls_in_row(); + else + covering_null_row_width= 0; + + if (covering_null_row_width) + count_pm_keys= nn_key_parts ? 1 : 0; + else + count_pm_keys= count_partial_match_columns - count_null_only_columns + + (nn_key_parts ? 1 : 0); + + choose_partial_match_strategy(test(nn_key_parts), + test(covering_null_row_width), + &partial_match_key_parts); + DBUG_ASSERT(strategy == PARTIAL_MATCH_MERGE || + strategy == PARTIAL_MATCH_SCAN); + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_engine= + new subselect_rowid_merge_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + count_pm_keys, + covering_null_row_width, + item, result, + semi_join_conds->argument_list()); + if (!pm_engine || + ((subselect_rowid_merge_engine*) pm_engine)-> + init(nn_key_parts, &partial_match_key_parts)) + { + /* + The call to init() would fail if there was not enough memory to allocate + all buffers for the rowid merge strategy. In this case revert to table + scanning which doesn't need any big buffers. + */ + delete pm_engine; + pm_engine= NULL; + strategy= PARTIAL_MATCH_SCAN; + } + } + + if (strategy == PARTIAL_MATCH_SCAN) + { + if (!(pm_engine= + new subselect_table_scan_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + item, result, + semi_join_conds->argument_list(), + covering_null_row_width))) + { + /* This is an irrecoverable error. */ + res= 1; + goto err; + } + } + } + + if (pm_engine) + lookup_engine= pm_engine; + item_in->change_engine(lookup_engine); err: - thd->lex->current_select= save_select; - if (res) - DBUG_RETURN(res); - } - - /* - Lookup the left IN operand in the hash index of the materialized subquery. - */ - DBUG_RETURN(subselect_uniquesubquery_engine::exec()); + thd->lex->current_select= save_select; + DBUG_RETURN(res); } @@ -3551,10 +4111,1008 @@ str->append(STRING_WITH_LEN(" <materialize> (")); materialize_engine->print(str, query_type); str->append(STRING_WITH_LEN(" ), ")); - if (tab) - subselect_uniquesubquery_engine::print(str, query_type); + + if (lookup_engine) + lookup_engine->print(str, query_type); else str->append(STRING_WITH_LEN( - "<the access method for lookups is not yet created>" + "<engine selected at execution time>" )); } + +void subselect_hash_sj_engine::fix_length_and_dec(Item_cache** row) +{ + DBUG_ASSERT(FALSE); +} + +void subselect_hash_sj_engine::exclude() +{ + DBUG_ASSERT(FALSE); +} + +bool subselect_hash_sj_engine::no_tables() +{ + DBUG_ASSERT(FALSE); + return FALSE; +} + +bool subselect_hash_sj_engine::change_result(Item_subselect *si, + select_result_interceptor *res) +{ + DBUG_ASSERT(FALSE); + return TRUE; +} + + +Ordered_key::Ordered_key(uint keyid_arg, TABLE *tbl_arg, Item *search_key_arg, + ha_rows null_count_arg, ha_rows min_null_row_arg, + ha_rows max_null_row_arg, uchar *row_num_to_rowid_arg) + : keyid(keyid_arg), tbl(tbl_arg), search_key(search_key_arg), + row_num_to_rowid(row_num_to_rowid_arg), null_count(null_count_arg) +{ + DBUG_ASSERT(tbl->file->stats.records > null_count); + key_buff_elements= tbl->file->stats.records - null_count; + cur_key_idx= HA_POS_ERROR; + + DBUG_ASSERT((null_count && min_null_row_arg && max_null_row_arg) || + (!null_count && !min_null_row_arg && !max_null_row_arg)); + if (null_count) + { + /* The counters are 1-based, for key access we need 0-based indexes. */ + min_null_row= min_null_row_arg - 1; + max_null_row= max_null_row_arg - 1; + } + else + min_null_row= max_null_row= 0; +} + + +Ordered_key::~Ordered_key() +{ + my_free((char*) key_buff, MYF(0)); + bitmap_free(&null_key); +} + + +/* + Cleanup that needs to be done for each PS (re)execution. +*/ + +void Ordered_key::cleanup() +{ + /* + Currently these keys are recreated for each PS re-execution, thus + there is nothing to cleanup, the whole object goes away after execution + is over. All handler related initialization/deinitialization is done by + the parent subselect_rowid_merge_engine object. + */ +} + + +/* + Initialize a multi-column index. +*/ + +bool Ordered_key::init(MY_BITMAP *columns_to_index) +{ + THD *thd= tbl->in_use; + uint cur_key_col= 0; + Item_field *cur_tmp_field; + Item_func_lt *fn_less_than; + + key_column_count= bitmap_bits_set(columns_to_index); + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(key_column_count * + sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(key_column_count * + sizeof(Item_func_lt*)); + + for (uint i= 0; i < columns_to_index->n_bits; i++) + { + if (!bitmap_is_set(columns_to_index, i)) + continue; + cur_tmp_field= new Item_field(tbl->field[i]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + fn_less_than= new Item_func_lt(cur_tmp_field, + search_key->element_index(i)); + fn_less_than->fix_fields(thd, (Item**) &fn_less_than); + key_columns[cur_key_col]= cur_tmp_field; + compare_pred[cur_key_col]= fn_less_than; + ++cur_key_col; + } + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Initialize a single-column index. +*/ + +bool Ordered_key::init(int col_idx) +{ + THD *thd= tbl->in_use; + + key_column_count= 1; + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(sizeof(Item_func_lt*)); + + key_columns[0]= new Item_field(tbl->field[col_idx]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + compare_pred[0]= new Item_func_lt(key_columns[0], + search_key->element_index(col_idx)); + compare_pred[0]->fix_fields(thd, (Item**)&compare_pred[0]); + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Allocate the buffers for both the row number, and the NULL-bitmap indexes. +*/ + +bool Ordered_key::alloc_keys_buffers() +{ + DBUG_ASSERT(key_buff_elements > 0); + + if (!(key_buff= (rownum_t*) my_malloc(key_buff_elements * sizeof(rownum_t), + MYF(MY_WME)))) + return TRUE; + + /* + TIMOUR: it is enough to create bitmaps with size + (max_null_row - min_null_row), and then use min_null_row as + lookup offset. + */ + /* Notice that max_null_row is max array index, we need count, so +1. */ + if (bitmap_init(&null_key, NULL, max_null_row + 1, FALSE)) + return TRUE; + + cur_key_idx= HA_POS_ERROR; + + return FALSE; +} + + +/* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + + @retval -1 + @retval 0 + @retval +1 +*/ + +int +Ordered_key::cmp_keys_by_row_data(ha_rows a, ha_rows b) +{ + uchar *rowid_a, *rowid_b; + int error, cmp_res; + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + + if (a == b) + return 0; + /* Get the corresponding rowids. */ + rowid_a= row_num_to_rowid + a * rowid_length; + rowid_b= row_num_to_rowid + b * rowid_length; + /* Fetch the rows for comparison. */ + error= tbl->file->ha_rnd_pos(tbl->record[0], rowid_a); + DBUG_ASSERT(!error); + error= tbl->file->ha_rnd_pos(tbl->record[1], rowid_b); + DBUG_ASSERT(!error); + /* + Compare the two rows by the corresponding values of the indexed + columns. + */ + for (uint i= 0; i < key_column_count; i++) + { + Field *cur_field= key_columns[i]->field; + if ((cmp_res= cur_field->cmp_offset(tbl->s->rec_buff_length))) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +int +Ordered_key::cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b) +{ + /* The result of comparing the two keys according to their row data. */ + int cmp_row_res= key->cmp_keys_by_row_data(*a, *b); + if (cmp_row_res) + return cmp_row_res; + return (*a < *b) ? -1 : (*a > *b) ? 1 : 0; +} + + +void Ordered_key::sort_keys() +{ + my_qsort2(key_buff, key_buff_elements, sizeof(rownum_t), + (qsort2_cmp) &cmp_keys_by_row_data_and_rownum, (void*) this); + /* Invalidate the current row position. */ + cur_key_idx= HA_POS_ERROR; +} + + +/* + The fraction of rows that do not contain NULL in the columns indexed by + this key. + + @retval 1 if there are no NULLs + @retval 0 if only NULLs +*/ + +double Ordered_key::null_selectivity() +{ + /* We should not be processing empty tables. */ + DBUG_ASSERT(tbl->file->stats.records); + return (1 - (double) null_count / (double) tbl->file->stats.records); +} + + +/* + Compare the value(s) of the current key in 'search_key' with the + data of the current table record. + + @notes The comparison result follows from the way compare_pred + is created in Ordered_key::init. Currently compare_pred compares + a field in of the current row with the corresponding Item that + contains the search key. + + @param row_num Number of the row (not index in the key_buff array) + + @retval -1 if (current row < search_key) + @retval 0 if (current row == search_key) + @retval +1 if (current row > search_key) +*/ + +int Ordered_key::cmp_key_with_search_key(rownum_t row_num) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + uchar *cur_rowid= row_num_to_rowid + row_num * rowid_length; + int error, cmp_res; + + error= tbl->file->ha_rnd_pos(tbl->record[0], cur_rowid); + DBUG_ASSERT(!error); + + for (uint i= 0; i < key_column_count; i++) + { + cmp_res= compare_pred[i]->get_comparator()->compare(); + /* Unlike Arg_comparator::compare_row() here there should be no NULLs. */ + DBUG_ASSERT(!compare_pred[i]->null_value); + if (cmp_res) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +/* + Find a key in a sorted array of keys via binary search. + + see create_subq_in_equalities() +*/ + +bool Ordered_key::lookup() +{ + DBUG_ASSERT(key_buff_elements); + + ha_rows lo= 0; + ha_rows hi= key_buff_elements - 1; + ha_rows mid; + int cmp_res; + + while (lo <= hi) + { + mid= lo + (hi - lo) / 2; + cmp_res= cmp_key_with_search_key(key_buff[mid]); + /* + In order to find the minimum match, check if the pevious element is + equal or smaller than the found one. If equal, we need to search further + to the left. + */ + if (!cmp_res && mid > 0) + cmp_res= !cmp_key_with_search_key(key_buff[mid - 1]) ? 1 : 0; + + if (cmp_res == -1) + { + /* row[mid] < search_key */ + lo= mid + 1; + } + else if (cmp_res == 1) + { + /* row[mid] > search_key */ + if (!mid) + goto not_found; + hi= mid - 1; + } + else + { + /* row[mid] == search_key */ + cur_key_idx= mid; + return TRUE; + } + } +not_found: + cur_key_idx= HA_POS_ERROR; + return FALSE; +} + + +/* + Move the current index pointer to the next key with the same column + values as the current key. Since the index is sorted, all such keys + are contiguous. +*/ + +bool Ordered_key::next_same() +{ + DBUG_ASSERT(key_buff_elements); + + if (cur_key_idx < key_buff_elements - 1) + { + /* + TIMOUR: + The below is quite inefficient, since as a result we will fetch every + row (except the last one) twice. There must be a more efficient way, + e.g. swapping record[0] and record[1], and reading only the new record. + */ + if (!cmp_keys_by_row_data(key_buff[cur_key_idx], key_buff[cur_key_idx + 1])) + { + ++cur_key_idx; + return TRUE; + } + } + return FALSE; +} + + +void Ordered_key::print(String *str) +{ + uint i; + str->append("{idx="); + str->qs_append(keyid); + str->append(", ("); + for (i= 0; i < key_column_count - 1; i++) + { + str->append(key_columns[i]->field->field_name); + str->append(", "); + } + str->append(key_columns[i]->field->field_name); + str->append("), "); + + str->append("null_bitmap: (bits="); + str->qs_append(null_key.n_bits); + str->append(", nulls= "); + str->qs_append((double)null_count); + str->append(", min_null= "); + str->qs_append((double)min_null_row); + str->append(", max_null= "); + str->qs_append((double)max_null_row); + str->append("), "); + + str->append('}'); +} + + +subselect_partial_match_engine::subselect_partial_match_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_engine(item_arg, result_arg), + tmp_table(tmp_table_arg), lookup_engine(engine_arg), + equi_join_conds(equi_join_conds_arg), + covering_null_row_width(covering_null_row_width_arg) +{} + + +int subselect_partial_match_engine::exec() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + int res; + + /* Try to find a matching row by index lookup. */ + res= lookup_engine->copy_ref_key_simple(); + if (res == -1) + { + /* The result is FALSE based on the outer reference. */ + item_in->value= 0; + item_in->null_value= 0; + return 0; + } + else if (res == 0) + { + /* Search for a complete match. */ + if ((res= lookup_engine->index_lookup())) + { + /* An error occured during lookup(). */ + item_in->value= 0; + item_in->null_value= 0; + return res; + } + else if (item_in->value) + { + /* + A complete match was found, the result of IN is TRUE. + Notice: (this->item == lookup_engine->item) + */ + return 0; + } + } + + if (covering_null_row_width == tmp_table->s->fields) + { + /* + If there is a NULL-only row that coveres all columns the result of IN + is UNKNOWN. + */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + return 0; + } + + /* + There is no complete match. Look for a partial match (UNKNOWN result), or + no match (FALSE). + */ + if (tmp_table->file->inited) + tmp_table->file->ha_index_end(); + + if (partial_match()) + { + /* The result of IN is UNKNOWN. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + } + else + { + /* The result of IN is FALSE. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 0; + item_in->null_value= 0; + } + + return 0; +} + + +void subselect_partial_match_engine::print(String *str, + enum_query_type query_type) +{ + /* + Should never be called as the actual engine cannot be known at query + optimization time. + */ + DBUG_ASSERT(FALSE); +} + + +/* + @param non_null_key_parts + @param partial_match_key_parts A union of all single-column NULL key parts. + @param count_partial_match_columns Number of NULL keyparts (set bits above). + + @retval FALSE the engine was initialized successfully + @retval TRUE there was some (memory allocation) error during initialization, + such errors should be interpreted as revert to other strategy +*/ + +bool +subselect_rowid_merge_engine::init(MY_BITMAP *non_null_key_parts, + MY_BITMAP *partial_match_key_parts) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tmp_table->file->ref_length; + ha_rows row_count= tmp_table->file->stats.records; + rownum_t cur_rownum= 0; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + uint cur_keyid= 0; + Item_in_subselect *item_in= (Item_in_subselect*) item; + int error; + + if (keys_count == 0) + { + /* There is nothing to initialize, we will only do regular lookups. */ + return FALSE; + } + + DBUG_ASSERT(!covering_null_row_width || (covering_null_row_width && + keys_count == 1 && + non_null_key_parts)); + /* + Allocate buffers to hold the merged keys and the mapping between rowids and + row numbers. + */ + if (!(merge_keys= (Ordered_key**) thd->alloc(keys_count * + sizeof(Ordered_key*))) || + !(row_num_to_rowid= (uchar*) my_malloc(row_count * rowid_length * + sizeof(uchar), MYF(MY_WME)))) + return TRUE; + + /* Create the only non-NULL key if there is any. */ + if (non_null_key_parts) + { + non_null_key= new Ordered_key(cur_keyid, tmp_table, item_in->left_expr, + 0, 0, 0, row_num_to_rowid); + if (non_null_key->init(non_null_key_parts)) + return TRUE; + merge_keys[cur_keyid]= non_null_key; + merge_keys[cur_keyid]->first(); + ++cur_keyid; + } + + /* + If there is a covering NULL row, the only key that is needed is the + only non-NULL key that is already created above. We create keys on + NULL-able columns only if there is no covering NULL row. + */ + if (!covering_null_row_width) + { + if (bitmap_init_memroot(&matching_keys, keys_count, thd->mem_root) || + bitmap_init_memroot(&matching_outer_cols, keys_count, thd->mem_root) || + bitmap_init_memroot(&null_only_columns, keys_count, thd->mem_root)) + return TRUE; + + /* + Create one single-column NULL-key for each column in + partial_match_key_parts. + */ + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i)) + continue; + + if (result_sink->get_null_count_of_col(i) == row_count) + bitmap_set_bit(&null_only_columns, cur_keyid); + else + { + merge_keys[cur_keyid]= new Ordered_key( + cur_keyid, tmp_table, + item_in->left_expr->element_index(i), + result_sink->get_null_count_of_col(i), + result_sink->get_min_null_of_col(i), + result_sink->get_max_null_of_col(i), + row_num_to_rowid); + if (merge_keys[cur_keyid]->init(i)) + return TRUE; + merge_keys[cur_keyid]->first(); + } + ++cur_keyid; + } + } + + /* Populate the indexes with data from the temporary table. */ + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + tmp_table->null_row= 0; + while (TRUE) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error == HA_ERR_RECORD_DELETED) + { + /* We get this for duplicate records that should not be in tmp_table. */ + continue; + } + /* + This is a temp table that we fully own, there should be no other + cause to stop the iteration than EOF. + */ + DBUG_ASSERT(!error || error == HA_ERR_END_OF_FILE); + if (error == HA_ERR_END_OF_FILE) + { + DBUG_ASSERT(cur_rownum == tmp_table->file->stats.records); + break; + } + + /* + Save the position of this record in the row_num -> rowid mapping. + */ + tmp_table->file->position(tmp_table->record[0]); + memcpy(row_num_to_rowid + cur_rownum * rowid_length, + tmp_table->file->ref, rowid_length); + + /* Add the current row number to the corresponding keys. */ + if (non_null_key) + { + /* By definition there are no NULLs in the non-NULL key. */ + non_null_key->add_key(cur_rownum); + } + + for (uint i= (non_null_key ? 1 : 0); i < keys_count; i++) + { + /* + Check if the first and only indexed column contains NULL in the curent + row, and add the row number to the corresponding key. + */ + if (tmp_table->field[merge_keys[i]->get_field_idx(0)]->is_null()) + merge_keys[i]->set_null(cur_rownum); + else + merge_keys[i]->add_key(cur_rownum); + } + ++cur_rownum; + } + + tmp_table->file->ha_rnd_end(); + + /* Sort all the keys by their NULL selectivity. */ + my_qsort(merge_keys, keys_count, sizeof(Ordered_key*), + (qsort_cmp) cmp_keys_by_null_selectivity); + + /* Sort the keys in each of the indexes. */ + for (uint i= 0; i < keys_count; i++) + merge_keys[i]->sort_keys(); + + if (init_queue(&pq, keys_count, 0, FALSE, + subselect_rowid_merge_engine::cmp_keys_by_cur_rownum, NULL)) + return TRUE; + + return FALSE; +} + + +subselect_rowid_merge_engine::~subselect_rowid_merge_engine() +{ + /* None of the resources below is allocated if there are no ordered keys. */ + if (keys_count) + { + my_free((char*) row_num_to_rowid, MYF(0)); + for (uint i= 0; i < keys_count; i++) + delete merge_keys[i]; + delete_queue(&pq); + if (tmp_table->file->inited == handler::RND) + tmp_table->file->ha_rnd_end(); + } +} + + +void subselect_rowid_merge_engine::cleanup() +{ +} + + +/* + Quick sort comparison function to compare keys in order of decreasing bitmap + selectivity, so that the most selective keys come first. + + @param k1 first key to compare + @param k2 second key to compare + + @retval 1 if k1 is less selective than k2 + @retval 0 if k1 is equally selective as k2 + @retval -1 if k1 is more selective than k2 +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_null_selectivity(Ordered_key **k1, + Ordered_key **k2) +{ + double k1_sel= (*k1)->null_selectivity(); + double k2_sel= (*k2)->null_selectivity(); + if (k1_sel < k2_sel) + return 1; + if (k1_sel > k2_sel) + return -1; + return 0; +} + + +/* +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_cur_rownum(void *arg, + uchar *k1, uchar *k2) +{ + rownum_t r1= ((Ordered_key*) k1)->current(); + rownum_t r2= ((Ordered_key*) k2)->current(); + + return (r1 < r2) ? -1 : (r1 > r2) ? 1 : 0; +} + + +/* + Check if certain table row contains a NULL in all columns for which there is + no match in the corresponding value index. + + @retval TRUE if a NULL row exists + @retval FALSE otherwise +*/ + +bool subselect_rowid_merge_engine::test_null_row(rownum_t row_num) +{ + Ordered_key *cur_key; + uint cur_id; + for (uint i = 0; i < keys_count; i++) + { + cur_key= merge_keys[i]; + cur_id= cur_key->get_keyid(); + if (bitmap_is_set(&matching_keys, cur_id)) + { + /* + The key 'i' (with id 'cur_keyid') already matches a value in row 'row_num', + thus we skip it as it can't possibly match a NULL. + */ + continue; + } + if (!cur_key->is_null(row_num)) + return FALSE; + } + return TRUE; +} + + +/* + @retval TRUE there is a partial match (UNKNOWN) + @retval FALSE there is no match at all (FALSE) +*/ + +bool subselect_rowid_merge_engine::partial_match() +{ + Ordered_key *min_key; /* Key that contains the current minimum position. */ + rownum_t min_row_num; /* Current row number of min_key. */ + Ordered_key *cur_key; + rownum_t cur_row_num; + uint count_nulls_in_search_key= 0; + bool res= FALSE; + + /* If there is a non-NULL key, it must be the first key in the keys array. */ + DBUG_ASSERT(!non_null_key || (non_null_key && merge_keys[0] == non_null_key)); + + /* All data accesses during execution are via handler::ha_rnd_pos() */ + tmp_table->file->ha_rnd_init(0); + + /* Check if there is a match for the columns of the only non-NULL key. */ + if (non_null_key && !non_null_key->lookup()) + { + res= FALSE; + goto end; + } + + /* + If there is a NULL (sub)row that covers all NULL-able columns, + then there is a guranteed partial match, and we don't need to search + for the matching row. + */ + if (covering_null_row_width) + { + res= TRUE; + goto end; + } + + if (non_null_key) + queue_insert(&pq, (uchar *) non_null_key); + /* + Do not add the non_null_key, since it was already processed above. + */ + bitmap_clear_all(&matching_outer_cols); + for (uint i= test(non_null_key); i < keys_count; i++) + { + DBUG_ASSERT(merge_keys[i]->get_column_count() == 1); + if (merge_keys[i]->get_search_key(0)->is_null()) + { + ++count_nulls_in_search_key; + bitmap_set_bit(&matching_outer_cols, merge_keys[i]->get_keyid()); + } + else if (merge_keys[i]->lookup()) + queue_insert(&pq, (uchar *) merge_keys[i]); + } + + /* + If the outer reference consists of only NULLs, or if it has NULLs in all + nullable columns, the result is UNKNOWN. + */ + if (count_nulls_in_search_key == + ((Item_in_subselect *) item)->left_expr->cols() - + (non_null_key ? non_null_key->get_column_count() : 0)) + { + res= TRUE; + goto end; + } + + /* + If there is no NULL (sub)row that covers all NULL columns, and there is no + single match for any of the NULL columns, the result is FALSE. + */ + if (pq.elements - test(non_null_key) == 0) + { + res= FALSE; + goto end; + } + + DBUG_ASSERT(pq.elements); + + min_key= (Ordered_key*) queue_remove(&pq, 0); + min_row_num= min_key->current(); + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + if (min_key->next_same()) + queue_insert(&pq, (uchar *) min_key); + + if (pq.elements == 0) + { + /* + Check the only matching row of the only key min_key for NULL matches + in the other columns. + */ + res= test_null_row(min_row_num); + goto end; + } + + while (TRUE) + { + cur_key= (Ordered_key*) queue_remove(&pq, 0); + cur_row_num= cur_key->current(); + + if (cur_row_num == min_row_num) + bitmap_set_bit(&matching_keys, cur_key->get_keyid()); + else + { + /* Follows from the correct use of priority queue. */ + DBUG_ASSERT(cur_row_num > min_row_num); + if (test_null_row(min_row_num)) + { + res= TRUE; + goto end; + } + else + { + min_key= cur_key; + min_row_num= cur_row_num; + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + } + } + + if (cur_key->next_same()) + queue_insert(&pq, (uchar *) cur_key); + + if (pq.elements == 0) + { + /* Check the last row of the last column in PQ for NULL matches. */ + res= test_null_row(min_row_num); + goto end; + } + } + + /* We should never get here - all branches must be handled explicitly above. */ + DBUG_ASSERT(FALSE); + +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +subselect_table_scan_engine::subselect_table_scan_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg) +{} + + +/* + TIMOUR: + This method is based on subselect_uniquesubquery_engine::scan_table(). + Consider refactoring somehow, 80% of the code is the same. + + for each row_i in tmp_table + { + count_matches= 0; + for each row element row_i[j] + { + if (outer_ref[j] is NULL || row_i[j] is NULL || outer_ref[j] == row_i[j]) + ++count_matches; + } + if (count_matches == outer_ref.elements) + return TRUE + } + return FALSE +*/ + +bool subselect_table_scan_engine::partial_match() +{ + List_iterator_fast<Item> equality_it(*equi_join_conds); + Item *cur_eq; + uint count_matches; + int error; + bool res; + + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + /* + TIMOUR: + scan_table() also calls "table->null_row= 0;", why, do we need it? + */ + for (;;) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(tmp_table, error); + break; + } + } + + equality_it.rewind(); + count_matches= 0; + while ((cur_eq= equality_it++)) + { + DBUG_ASSERT(cur_eq->type() == Item::FUNC_ITEM && + ((Item_func*)cur_eq)->functype() == Item_func::EQ_FUNC); + if (!cur_eq->val_int() && !cur_eq->null_value) + break; + ++count_matches; + } + if (count_matches == tmp_table->s->fields) + { + res= TRUE; /* Found a matching row. */ + goto end; + } + } + + res= FALSE; +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +void subselect_table_scan_engine::cleanup() +{ +} === modified file 'sql/item_subselect.h' --- a/sql/item_subselect.h 2010-02-11 23:59:58 +0000 +++ b/sql/item_subselect.h 2010-03-09 10:14:06 +0000 @@ -297,7 +297,7 @@ Representation of IN subquery predicates of the form "left_expr IN (SELECT ...)". - @detail + @details This class has: - A "subquery execution engine" (as a subclass of Item_subselect) that allows it to evaluate subqueries. (and this class participates in execution by @@ -319,6 +319,12 @@ */ List<Cached_item> *left_expr_cache; bool first_execution; + /* + Set to TRUE if at query execution time we determine that this item's + value is a constant during this execution. We need this member because + it is not possible to substitute 'this' with a constant item. + */ + bool is_constant; /* expr & optimizer used in subselect rewriting to store Item for @@ -387,8 +393,8 @@ Item_in_subselect(Item * left_expr, st_select_lex *select_lex); Item_in_subselect() :Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), abort_on_null(0), pushed_cond_guards(NULL), - exec_method(NOT_TRANSFORMED), upper_item(0) + is_constant(FALSE), optimizer(0), abort_on_null(0), + pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), upper_item(0) {} void cleanup(); subs_type substype() { return IN_SUBS; } @@ -421,6 +427,8 @@ void update_used_tables(); bool setup_engine(); bool init_left_expr_cache(); + /* Inform 'this' that it was computed, and contains a valid result. */ + void set_first_execution() { if (first_execution) first_execution= FALSE; } bool is_expensive_processor(uchar *arg); friend class Item_ref_null_helper; @@ -428,6 +436,7 @@ friend class Item_in_optimizer; friend class subselect_indexsubquery_engine; friend class subselect_hash_sj_engine; + friend class subselect_partial_match_engine; }; @@ -462,7 +471,8 @@ enum enum_engine_type {ABSTRACT_ENGINE, SINGLE_SELECT_ENGINE, UNION_ENGINE, UNIQUESUBQUERY_ENGINE, - INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE}; + INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE, + ROWID_MERGE_ENGINE, TABLE_SCAN_ENGINE}; subselect_engine(Item_subselect *si, select_result_interceptor *res) :thd(0) @@ -635,8 +645,10 @@ virtual void print (String *str, enum_query_type query_type); bool change_result(Item_subselect *si, select_result_interceptor *result); bool no_tables(); + int index_lookup(); /* TIMOUR: this method needs refactoring. */ int scan_table(); bool copy_ref_key(); + int copy_ref_key_simple(); /* TIMOUR: this method needs refactoring. */ bool no_rows() { return empty_result_set; } virtual enum_engine_type engine_type() { return UNIQUESUBQUERY_ENGINE; } }; @@ -705,50 +717,439 @@ /** - Compute an IN predicate via a hash semi-join. The subquery is materialized - during the first evaluation of the IN predicate. The IN predicate is executed - via the functionality inherited from subselect_uniquesubquery_engine. + Compute an IN predicate via a hash semi-join. This class is responsible for + the materialization of the subquery, and the selection of the correct and + optimal execution method (e.g. direct index lookup, or partial matching) for + the IN predicate. */ -class subselect_hash_sj_engine: public subselect_uniquesubquery_engine +class subselect_hash_sj_engine : public subselect_engine { protected: + /* The table into which the subquery is materialized. */ + TABLE *tmp_table; /* TRUE if the subquery was materialized into a temp table. */ bool is_materialized; /* The old engine already chosen at parse time and stored in permanent memory. Through this member we can re-create and re-prepare materialize_join for - each execution of a prepared statement. We akso resuse the functionality + each execution of a prepared statement. We also reuse the functionality of subselect_single_select_engine::[prepare | cols]. */ subselect_single_select_engine *materialize_engine; + /* The engine used to compute the IN predicate. */ + subselect_engine *lookup_engine; /* QEP to execute the subquery and materialize its result into a temporary table. Created during the first call to exec(). */ JOIN *materialize_join; - /* Temp table context of the outer select's JOIN. */ - TMP_TABLE_PARAM *tmp_param; + + /* Keyparts of the only non-NULL composite index in a rowid merge. */ + MY_BITMAP non_null_key_parts; + /* Keyparts of the single column indexes with NULL, one keypart per index. */ + MY_BITMAP partial_match_key_parts; + uint count_partial_match_columns; + uint count_null_only_columns; + /* + A conjunction of all the equality condtions between all pairs of expressions + that are arguments of an IN predicate. We need these to post-filter some + IN results because index lookups sometimes match values that are actually + not equal to the search key in SQL terms. + */ + Item_cond_and *semi_join_conds; + /* Possible execution strategies that can be used to compute hash semi-join.*/ + enum exec_strategy { + UNDEFINED, + COMPLETE_MATCH, /* Use regular index lookups. */ + PARTIAL_MATCH, /* Use some partial matching strategy. */ + PARTIAL_MATCH_MERGE, /* Use partial matching through index merging. */ + PARTIAL_MATCH_SCAN, /* Use partial matching through table scan. */ + IMPOSSIBLE /* Subquery materialization is not applicable. */ + }; + /* The chosen execution strategy. Computed after materialization. */ + exec_strategy strategy; +protected: + exec_strategy get_strategy_using_schema(); + exec_strategy get_strategy_using_data(); + size_t rowid_merge_buff_size(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + void choose_partial_match_strategy(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + bool make_semi_join_conds(); + subselect_uniquesubquery_engine* make_unique_engine(); public: subselect_hash_sj_engine(THD *thd, Item_subselect *in_predicate, - subselect_single_select_engine *old_engine) - :subselect_uniquesubquery_engine(thd, NULL, in_predicate, NULL), - is_materialized(FALSE), materialize_engine(old_engine), - materialize_join(NULL), tmp_param(NULL) - {} + subselect_single_select_engine *old_engine) + :subselect_engine(in_predicate, NULL), tmp_table(NULL), + is_materialized(FALSE), materialize_engine(old_engine), lookup_engine(NULL), + materialize_join(NULL), count_partial_match_columns(0), + count_null_only_columns(0), semi_join_conds(NULL), strategy(UNDEFINED) + { + set_thd(thd); + } ~subselect_hash_sj_engine(); bool init_permanent(List<Item> *tmp_columns); bool init_runtime(); void cleanup(); - int prepare() { return 0; } + int prepare() { return 0; } /* Override virtual function in base class. */ int exec(); - virtual void print (String *str, enum_query_type query_type); + virtual void print(String *str, enum_query_type query_type); uint cols() { return materialize_engine->cols(); } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + table_map upper_select_const_tables() { return 0; } + bool no_rows() { return !tmp_table->file->stats.records; } virtual enum_engine_type engine_type() { return HASH_SJ_ENGINE; } -}; - + /* + TODO: factor out all these methods in a base subselect_index_engine class + because all of them have dummy implementations and should never be called. + */ + void fix_length_and_dec(Item_cache** row);//=>base class + void exclude(); //=>base class + //=>base class + bool change_result(Item_subselect *si, select_result_interceptor *result); + bool no_tables();//=>base class +}; + + +/* + Distinguish the type od (0-based) row numbers from the type of the index into + an array of row numbers. +*/ +typedef ha_rows rownum_t; + + +/* + An Ordered_key is an in-memory table index that allows O(log(N)) time + lookups of a multi-part key. + + If the index is over a single column, then this column may contain NULLs, and + the NULLs are stored and tested separately for NULL in O(1) via is_null(). + Multi-part indexes assume that the indexed columns do not contain NULLs. + + TODO: + = Due to the unnatural assymetry between single and multi-part indexes, it + makes sense to somehow refactor or extend the class. + + = This class can be refactored into a base abstract interface, and two + subclasses: + - one to represent single-column indexes, and + - another to represent multi-column indexes. + Such separation would allow slightly more efficient implementation of + the single-column indexes. + = The current design requires such indexes to be fully recreated for each + PS (re)execution, however most of the comprising objects can be reused. +*/ + +class Ordered_key : public Sql_alloc +{ +protected: + /* + Index of the key in an array of keys. This index allows to + construct (sub)sets of keys represented by bitmaps. + */ + uint keyid; + /* The table being indexed. */ + TABLE *tbl; + /* The columns being indexed. */ + Item_field **key_columns; + /* Number of elements in 'key_columns' (number of key parts). */ + uint key_column_count; + /* + An expression, or sequence of expressions that forms the search key. + The search key is a sequence when it is Item_row. Each element of the + sequence is accessible via Item::element_index(int i). + */ + Item *search_key; + +/* Value index related members. */ + /* + The actual value index, consists of a sorted sequence of row numbers. + */ + rownum_t *key_buff; + /* Number of elements in key_buff. */ + ha_rows key_buff_elements; + /* Current element in 'key_buff'. */ + ha_rows cur_key_idx; + /* + Mapping from row numbers to row ids. The element row_num_to_rowid[i] + contains a buffer with the rowid for the row numbered 'i'. + The memory for this member is not maintanined by this class because + all Ordered_key indexes of the same table share the same mapping. + */ + uchar *row_num_to_rowid; + /* + A sequence of predicates to compare the search key with the corresponding + columns of a table row from the index. + */ + Item_func_lt **compare_pred; + +/* Null index related members. */ + MY_BITMAP null_key; + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + +protected: + bool alloc_keys_buffers(); + /* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + */ + int cmp_keys_by_row_data(rownum_t a, rownum_t b); + static int cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b); + + int cmp_key_with_search_key(rownum_t row_num); + +public: + Ordered_key(uint keyid_arg, TABLE *tbl_arg, + Item *search_key_arg, ha_rows null_count_arg, + ha_rows min_null_row_arg, ha_rows max_null_row_arg, + uchar *row_num_to_rowid_arg); + ~Ordered_key(); + void cleanup(); + /* Initialize a multi-column index. */ + bool init(MY_BITMAP *columns_to_index); + /* Initialize a single-column index. */ + bool init(int col_idx); + + uint get_column_count() { return key_column_count; } + uint get_keyid() { return keyid; } + uint get_field_idx(uint i) + { + DBUG_ASSERT(i < key_column_count); + return key_columns[i]->field->field_index; + } + /* + Get the search key element that corresponds to the i-th key part of this + index. + */ + Item *get_search_key(uint i) + { + return search_key->element_index(key_columns[i]->field->field_index); + } + void add_key(rownum_t row_num) + { + /* The caller must know how many elements to add. */ + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + key_buff[cur_key_idx]= row_num; + ++cur_key_idx; + } + + void sort_keys(); + double null_selectivity(); + + /* + Position the current element at the first row that matches the key. + The key itself is propagated by evaluating the current value(s) of + this->search_key. + */ + bool lookup(); + /* Move the current index cursor to the first key. */ + void first() + { + DBUG_ASSERT(key_buff_elements); + cur_key_idx= 0; + } + /* TODO */ + bool next_same(); + /* Move the current index cursor to the next key. */ + bool next() + { + DBUG_ASSERT(key_buff_elements); + if (cur_key_idx < key_buff_elements - 1) + { + ++cur_key_idx; + return TRUE; + } + return FALSE; + }; + /* Return the current index element. */ + rownum_t current() + { + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + return key_buff[cur_key_idx]; + } + + void set_null(rownum_t row_num) + { + bitmap_set_bit(&null_key, row_num); + } + bool is_null(rownum_t row_num) + { + /* + Indexes consisting of only NULLs do not have a bitmap buffer at all. + Their only initialized member is 'n_bits', which is equal to the number + of temp table rows. + */ + if (null_count == tbl->file->stats.records) + { + DBUG_ASSERT(tbl->file->stats.records == null_key.n_bits); + return TRUE; + } + if (row_num > max_null_row || row_num < min_null_row) + return FALSE; + return bitmap_is_set(&null_key, row_num); + } + void print(String *str); +}; + + +class subselect_partial_match_engine : public subselect_engine +{ +protected: + /* The temporary table that contains a materialized subquery. */ + TABLE *tmp_table; + /* + The engine used to check whether an IN predicate is TRUE or not. If not + TRUE, then subselect_rowid_merge_engine further distinguishes between + FALSE and UNKNOWN. + */ + subselect_uniquesubquery_engine *lookup_engine; + /* A list of equalities between each pair of IN operands. */ + List<Item> *equi_join_conds; + /* + If there is a row, such that all its NULL-able components are NULL, this + member is set to the number of covered columns. If there is no covering + row, then this is 0. + */ + uint covering_null_row_width; +protected: + virtual bool partial_match()= 0; +public: + subselect_partial_match_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + int prepare() { return 0; } + int exec(); + void fix_length_and_dec(Item_cache**) {} + uint cols() { /* TODO: what is the correct value? */ return 1; } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + void exclude() {} + table_map upper_select_const_tables() { return 0; } + bool change_result(Item_subselect*, select_result_interceptor*) + { DBUG_ASSERT(FALSE); return false; } + bool no_tables() { return false; } + bool no_rows() + { + /* + TODO: It is completely unclear what is the semantics of this + method. The current result is computed so that the call to no_rows() + from Item_in_optimizer::val_int() sets Item_in_optimizer::null_value + correctly. + */ + return !(((Item_in_subselect *) item)->null_value); + } + void print(String*, enum_query_type); + + friend void subselect_hash_sj_engine::cleanup(); +}; + + +class subselect_rowid_merge_engine: public subselect_partial_match_engine +{ +protected: + /* + Mapping from row numbers to row ids. The rowids are stored sequentially + in the array - rowid[i] is located in row_num_to_rowid + i * rowid_length. + */ + uchar *row_num_to_rowid; + /* + A subset of all the keys for which there is a match for the same row. + Used during execution. Computed for each outer reference + */ + MY_BITMAP matching_keys; + /* + The columns of the outer reference that are NULL. Computed for each + outer reference. + */ + MY_BITMAP matching_outer_cols; + /* + Columns that consist of only NULLs. Such columns match any value. + Computed once per query execution. + */ + MY_BITMAP null_only_columns; + /* + Indexes of row numbers, sorted by <column_value, row_number>. If an + index may contain NULLs, the NULLs are stored efficiently in a bitmap. + + The indexes are sorted by the selectivity of their NULL sub-indexes, the + one with the fewer NULLs is first. Thus, if there is any index on + non-NULL columns, it is contained in keys[0]. + */ + Ordered_key **merge_keys; + /* The number of elements in keys. */ + uint keys_count; + /* + An index on all non-NULL columns of 'tmp_table'. The index has the + logical form: <[v_i1 | ... | v_ik], rownum>. It allows to find the row + number where the columns c_i1,...,c1_k contain the values v_i1,...,v_ik. + If such an index exists, it is always the first element of 'keys'. + */ + Ordered_key *non_null_key; + /* + Priority queue of Ordered_key indexes, one per NULLable column. + This queue is used by the partial match algorithm in method exec(). + */ + QUEUE pq; +protected: + /* + Comparison function to compare keys in order of decreasing bitmap + selectivity. + */ + static int cmp_keys_by_null_selectivity(Ordered_key **k1, Ordered_key **k2); + /* + Comparison function used by the priority queue pq, the 'smaller' key + is the one with the smaller current row number. + */ + static int cmp_keys_by_cur_rownum(void *arg, uchar *k1, uchar *k2); + + bool test_null_row(rownum_t row_num); + bool partial_match(); +public: + subselect_rowid_merge_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, uint keys_count_arg, + uint covering_null_row_width_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg), + keys_count(keys_count_arg), non_null_key(NULL) + { + thd= lookup_engine->get_thd(); + } + ~subselect_rowid_merge_engine(); + bool init(MY_BITMAP *non_null_key_parts, MY_BITMAP *partial_match_key_parts); + void cleanup(); + virtual enum_engine_type engine_type() { return ROWID_MERGE_ENGINE; } +}; + + +class subselect_table_scan_engine: public subselect_partial_match_engine +{ +protected: + bool partial_match(); +public: + subselect_table_scan_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + void cleanup(); + virtual enum_engine_type engine_type() { return TABLE_SCAN_ENGINE; } +}; === modified file 'sql/mysql_priv.h' --- a/sql/mysql_priv.h 2010-01-17 14:55:08 +0000 +++ b/sql/mysql_priv.h 2010-03-09 10:14:06 +0000 @@ -552,12 +552,14 @@ #define OPTIMIZER_SWITCH_LOOSE_SCAN 64 #define OPTIMIZER_SWITCH_MATERIALIZATION 128 #define OPTIMIZER_SWITCH_SEMIJOIN 256 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE 512 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN 1024 #ifdef DBUG_OFF -# define OPTIMIZER_SWITCH_LAST 512 +# define OPTIMIZER_SWITCH_LAST 2048 #else -# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 512 -# define OPTIMIZER_SWITCH_LAST 1024 +# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 2048 +# define OPTIMIZER_SWITCH_LAST 4096 #endif #ifdef DBUG_OFF @@ -570,8 +572,10 @@ OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) -#else + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) +#else # define OPTIMIZER_SWITCH_DEFAULT (OPTIMIZER_SWITCH_INDEX_MERGE | \ OPTIMIZER_SWITCH_INDEX_MERGE_UNION | \ OPTIMIZER_SWITCH_INDEX_MERGE_SORT_UNION | \ @@ -581,7 +585,9 @@ OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) #endif /* === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2010-01-17 14:55:08 +0000 +++ b/sql/mysqld.cc 2010-03-09 10:14:06 +0000 @@ -301,7 +301,9 @@ "index_merge","index_merge_union","index_merge_sort_union", "index_merge_intersection", "index_condition_pushdown", - "firstmatch","loosescan","materialization", "semijoin", + "firstmatch","loosescan","materialization", "semijoin", + "partial_match_rowid_merge", + "partial_match_table_scan", #ifndef DBUG_OFF "table_elimination", #endif @@ -320,6 +322,8 @@ sizeof("loosescan") - 1, sizeof("materialization") - 1, sizeof("semijoin") - 1, + sizeof("partial_match_rowid_merge") - 1, + sizeof("partial_match_table_scan") - 1, #ifndef DBUG_OFF sizeof("table_elimination") - 1, #endif @@ -5794,7 +5798,8 @@ OPT_RECORD_RND_BUFFER, OPT_DIV_PRECINCREMENT, OPT_RELAY_LOG_SPACE_LIMIT, OPT_RELAY_LOG_PURGE, OPT_SLAVE_NET_TIMEOUT, OPT_SLAVE_COMPRESSED_PROTOCOL, OPT_SLOW_LAUNCH_TIME, - OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_DEBUGGING, OPT_DEBUG_FLUSH, + OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_ROWID_MERGE_BUFF_SIZE, + OPT_DEBUGGING, OPT_DEBUG_FLUSH, OPT_SORT_BUFFER, OPT_TABLE_OPEN_CACHE, OPT_TABLE_DEF_CACHE, OPT_THREAD_CONCURRENCY, OPT_THREAD_CACHE_SIZE, OPT_TMP_TABLE_SIZE, OPT_THREAD_STACK, @@ -7130,6 +7135,11 @@ (uchar**) &max_system_variables.range_alloc_block_size, 0, GET_ULONG, REQUIRED_ARG, RANGE_ALLOC_BLOCK_SIZE, RANGE_ALLOC_BLOCK_SIZE, (longlong) ULONG_MAX, 0, 1024, 0}, + {"rowid_merge_buff_size", OPT_ROWID_MERGE_BUFF_SIZE, + "The size of the buffers used [NOT] IN evaluation via partial matching.", + (uchar**) &global_system_variables.rowid_merge_buff_size, + (uchar**) &max_system_variables.rowid_merge_buff_size, 0, GET_ULONG, + REQUIRED_ARG, 8*1024*1024L, 0, MAX_MEM_TABLE_SIZE/2, 0, 1, 0}, {"read_buffer_size", OPT_RECORD_BUFFER, "Each thread that does a sequential scan allocates a buffer of this size for each table it scans. If you do many sequential scans, you may want to increase this value.", (uchar**) &global_system_variables.read_buff_size, === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-15 06:32:54 +0000 +++ b/sql/opt_subselect.cc 2010-03-15 19:52:58 +0000 @@ -187,10 +187,10 @@ does not call setup_subquery_materialization(). We could make SELECT ... FROM DUAL call that function but that doesn't seem to be the case that is worth handling. - 4. Subquery predicate is a top-level predicate - (this implies it is not negated) - TODO: this is a limitation that should be lifted once we - implement correct NULL semantics (WL#3830) + 4. Either the subquery predicate is a top-level predicate, or at + least one partial match strategy is enabled. If no partial match + strategy is enabled, then materialization cannot be used for + non-top-level queries because it cannot handle NULLs correctly. 5. Subquery is non-correlated TODO: This is an overly restrictive condition. It can be extended to: @@ -204,8 +204,8 @@ (*) The subquery must be part of a SELECT statement. The current condition also excludes multi-table update statements. - We have to determine whether we will perform subquery materialization - before calling the IN=>EXISTS transformation, so that we know whether to + Determine whether we will perform subquery materialization before + calling the IN=>EXISTS transformation, so that we know whether to perform the whole transformation or only that part of it which wraps Item_in_subselect in an Item_in_optimizer. */ @@ -215,12 +215,14 @@ select_lex->master_unit()->first_select()->leaf_tables && // 3 thd->lex->sql_command == SQLCOM_SELECT && // * select_lex->outer_select()->leaf_tables && // 3A - subquery_types_allow_materialization(in_subs)) + subquery_types_allow_materialization(in_subs) && + // psergey-todo: duplicated_subselect_card_check: where it's done? + (in_subs->is_top_level_item() || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) &&//4 + !in_subs->is_correlated && // 5 + in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 { - // psergey-todo: duplicated_subselect_card_check: where it's done? - if (in_subs->is_top_level_item() && // 4 - !in_subs->is_correlated && // 5 - in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 in_subs->exec_method= Item_in_subselect::MATERIALIZATION; } === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-12-22 12:49:15 +0000 +++ b/sql/set_var.cc 2010-03-09 10:14:06 +0000 @@ -540,6 +540,9 @@ static sys_var_thd_ulong sys_range_alloc_block_size(&vars, "range_alloc_block_size", &SV::range_alloc_block_size); +static sys_var_thd_ulong sys_rowid_merge_buff_size(&vars, "rowid_merge_buff_size", + &SV::rowid_merge_buff_size); + static sys_var_thd_ulong sys_query_alloc_block_size(&vars, "query_alloc_block_size", &SV::query_alloc_block_size, 0, fix_thd_mem_root); === modified file 'sql/sql_class.cc' --- a/sql/sql_class.cc 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.cc 2010-02-19 21:55:57 +0000 @@ -42,6 +42,7 @@ #include "sp_rcontext.h" #include "sp_cache.h" +#include "sql_select.h" /* declares create_tmp_table() */ /* The following is used to initialise Table_ident with a internal @@ -2877,6 +2878,71 @@ return 0; } + +bool +select_materialize_with_stats:: +create_result_table(THD *thd_arg, List<Item> *column_types, + bool is_union_distinct, ulonglong options, + const char *table_alias, bool bit_fields_as_long) +{ + DBUG_ASSERT(table == 0); + tmp_table_param.field_count= column_types->elements; + tmp_table_param.bit_fields_as_long= bit_fields_as_long; + + if (! (table= create_tmp_table(thd_arg, &tmp_table_param, *column_types, + (ORDER*) 0, is_union_distinct, 1, + options, HA_POS_ERROR, (char*) table_alias))) + return TRUE; + + col_stat= (Column_statistics*) table->in_use->alloc(table->s->fields * + sizeof(Column_statistics)); + if (!stat) + return TRUE; + + cleanup(); + + table->file->extra(HA_EXTRA_WRITE_CACHE); + table->file->extra(HA_EXTRA_IGNORE_DUP_KEY); + return FALSE; +} + + +/** + Override select_union::send_data to analyze each row for NULLs and to + update null_statistics before sending data to the client. + + @return TRUE if fatal error when sending data to the client + @return FALSE on success +*/ + +bool select_materialize_with_stats::send_data(List<Item> &items) +{ + List_iterator_fast<Item> item_it(items); + Item *cur_item; + Column_statistics *cur_col_stat= col_stat; + uint nulls_in_row= 0; + + ++count_rows; + + while ((cur_item= item_it++)) + { + if (cur_item->is_null()) + { + ++cur_col_stat->null_count; + cur_col_stat->max_null_row= count_rows; + if (!cur_col_stat->min_null_row) + cur_col_stat->min_null_row= count_rows; + ++nulls_in_row; + } + ++cur_col_stat; + } + if (nulls_in_row > max_nulls_in_row) + max_nulls_in_row= nulls_in_row; + + return select_union::send_data(items); +} + + /**************************************************************************** TMP_TABLE_PARAM ****************************************************************************/ === modified file 'sql/sql_class.h' --- a/sql/sql_class.h 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.h 2010-03-09 10:14:06 +0000 @@ -343,6 +343,8 @@ ulong mrr_buff_size; ulong div_precincrement; ulong sortbuff_size; + /* Total size of all buffers used by the subselect_rowid_merge_engine. */ + ulong rowid_merge_buff_size; ulong thread_handling; ulong tx_isolation; ulong completion_type; @@ -2740,19 +2742,20 @@ class select_union :public select_result_interceptor { +protected: TMP_TABLE_PARAM tmp_table_param; public: TABLE *table; - select_union() :table(0) {} + select_union() :table(0) { tmp_table_param.init(); } int prepare(List<Item> &list, SELECT_LEX_UNIT *u); bool send_data(List<Item> &items); bool send_eof(); bool flush(); - bool create_result_table(THD *thd, List<Item> *column_types, - bool is_distinct, ulonglong options, - const char *alias, bool bit_fields_as_long); + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); }; /* Base subselect interface class */ @@ -2776,6 +2779,74 @@ bool send_data(List<Item> &items); }; + +/* + This class specializes select_union to collect statistics about the + data stored in the temp table. Currently the class collects statistcs + about NULLs. +*/ + +class select_materialize_with_stats : public select_union +{ +protected: + class Column_statistics + { + public: + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + }; + + /* Array of statistics data per column. */ + Column_statistics* col_stat; + + /* + The number of columns in the biggest sub-row that consists of only + NULL values. + */ + ha_rows max_nulls_in_row; + /* + Count of rows writtent to the temp table. This is redundant as it is + already stored in handler::stats.records, however that one is relatively + expensive to compute (given we need that for evry row). + */ + ha_rows count_rows; + +public: + select_materialize_with_stats() {} + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); + bool init_result_table(ulonglong select_options); + bool send_data(List<Item> &items); + void cleanup() + { + memset(col_stat, 0, table->s->fields * sizeof(Column_statistics)); + max_nulls_in_row= 0; + count_rows= 0; + } + ha_rows get_null_count_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].null_count; + } + ha_rows get_max_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].max_null_row; + } + ha_rows get_min_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].min_null_row; + } + ha_rows get_max_nulls_in_row() { return max_nulls_in_row; } +}; + + /* used in independent ALL/ANY optimisation */ class select_max_min_finder_subselect :public select_subselect { === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-14 18:25:43 +0000 +++ b/sql/sql_select.cc 2010-03-15 19:52:58 +0000 @@ -874,6 +874,9 @@ { DBUG_PRINT("info",("No tables")); error= 0; + /* Create all structures needed for materialized subquery execution. */ + if (setup_subquery_materialization()) + DBUG_RETURN(1); DBUG_RETURN(0); } error= -1; // Error is sent to client @@ -11258,7 +11261,7 @@ param->group_buff=group_buff; share->keys=1; share->uniques= test(using_unique_constraint); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME; keyinfo->usable_key_parts=keyinfo->key_parts= param->group_parts; @@ -11344,7 +11347,7 @@ keyinfo->key_parts * sizeof(KEY_PART_INFO)))) goto err; bzero((void*) key_part_info, keyinfo->key_parts * sizeof(KEY_PART_INFO)); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME | HA_NULL_ARE_EQUAL; keyinfo->key_length= 0; // Will compute the sum of the parts below.

1 0

[Maria-developers] bzr commit into file:///home/tsk/mprog/src/5.3-subqueries/ branch (timour:2779)
by timour＠askmonty.org 15 Mar '10

15 Mar '10

#At file:///home/tsk/mprog/src/5.3-subqueries/ based on revid:psergey@askmonty.org-20100315063535-jsp4jgya6lfqt8e6 2779 timour(a)askmonty.org 2010-03-15 [merge] Merge in MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs modified: mysql-test/include/mix1.inc mysql-test/r/index_merge_myisam.result mysql-test/r/innodb_mysql.result mysql-test/r/myisam_mrr.result mysql-test/r/ps.result mysql-test/r/subselect.result mysql-test/r/subselect3.result mysql-test/r/subselect3_jcl6.result mysql-test/r/subselect_no_mat.result mysql-test/r/subselect_no_opts.result mysql-test/r/subselect_no_semijoin.result mysql-test/r/subselect_sj.result mysql-test/r/subselect_sj_jcl6.result mysql-test/t/ps.test mysql-test/t/subselect.test mysql-test/t/subselect3.test sql/item_cmpfunc.h sql/item_subselect.cc sql/item_subselect.h sql/mysql_priv.h sql/mysqld.cc sql/opt_subselect.cc sql/set_var.cc sql/sql_class.cc sql/sql_class.h sql/sql_select.cc === modified file 'mysql-test/include/mix1.inc' --- a/mysql-test/include/mix1.inc 2009-09-15 06:08:54 +0000 +++ b/mysql-test/include/mix1.inc 2010-03-11 21:43:31 +0000 @@ -1177,8 +1177,11 @@ DROP TABLE t1; create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; --echo End of 5.0 tests === modified file 'mysql-test/r/index_merge_myisam.result' --- a/mysql-test/r/index_merge_myisam.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/index_merge_myisam.result 2010-03-11 21:43:31 +0000 @@ -1419,19 +1419,19 @@ drop table t1; # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge=off,index_merge_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge_union=on'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,index_merge_sort_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=4; ERROR 42000: Variable 'optimizer_switch' can't be set to the value of '4' set optimizer_switch=NULL; @@ -1458,21 +1458,21 @@ set optimizer_switch=default; set optimizer_switch='index_merge=off,index_merge_union=off,default'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set @@global.optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on # # Check index_merge's @@optimizer_switch flags # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, c int, filler char(100), @@ -1582,5 +1582,5 @@ id select_type table type possible_keys set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on drop table t0, t1; === modified file 'mysql-test/r/innodb_mysql.result' --- a/mysql-test/r/innodb_mysql.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mysql.result 2010-03-11 21:43:31 +0000 @@ -1425,12 +1425,15 @@ DROP TABLE t1; # create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables 2 DEPENDENT SUBQUERY t1 system NULL NULL NULL NULL 0 const row not found 2 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 1 +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; End of 5.0 tests CREATE TABLE `t2` ( === modified file 'mysql-test/r/myisam_mrr.result' --- a/mysql-test/r/myisam_mrr.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/myisam_mrr.result 2010-03-11 21:43:31 +0000 @@ -394,7 +394,7 @@ drop table t0, t1; # - engine_condition_pushdown does not affect ICP select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, key(a)); === modified file 'mysql-test/r/ps.result' --- a/mysql-test/r/ps.result 2009-05-27 15:19:44 +0000 +++ b/mysql-test/r/ps.result 2010-03-11 21:43:31 +0000 @@ -149,6 +149,8 @@ c29 longblob, c30 longtext, c31 enum('on c32 set('monday', 'tuesday', 'wednesday') ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -177,6 +179,7 @@ id select_type table type possible_keys 2 DEPENDENT SUBQUERY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; set @arg00=1; prepare stmt1 from ' create table t1 (m int) as select 1 as m ' ; execute stmt1 ; === modified file 'mysql-test/r/subselect.result' --- a/mysql-test/r/subselect.result 2010-02-17 21:59:41 +0000 +++ b/mysql-test/r/subselect.result 2010-03-11 21:43:31 +0000 @@ -1,4 +1,6 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4803,4 +4805,5 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. === modified file 'mysql-test/r/subselect3.result' --- a/mysql-test/r/subselect3.result 2010-02-17 10:05:27 +0000 +++ b/mysql-test/r/subselect3.result 2010-03-11 21:43:31 +0000 @@ -63,12 +63,15 @@ Handler_read_rnd_next 11 select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -692,6 +695,8 @@ a MAX(b) test 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -759,6 +764,7 @@ id select_type table type possible_keys 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -960,7 +966,7 @@ i1 i2 # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -977,7 +983,7 @@ i1 i2 # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect3_jcl6.result' --- a/mysql-test/r/subselect3_jcl6.result 2010-02-17 10:47:55 +0000 +++ b/mysql-test/r/subselect3_jcl6.result 2010-03-11 21:43:31 +0000 @@ -67,12 +67,15 @@ Handler_read_rnd_next 11 select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -696,6 +699,8 @@ a MAX(b) test 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -763,6 +768,7 @@ id select_type table type possible_keys 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -964,7 +970,7 @@ i1 i2 # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -981,7 +987,7 @@ i1 i2 # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect_no_mat.result' --- a/mysql-test/r/subselect_no_mat.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_mat.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_opts.result' --- a/mysql-test/r/subselect_no_opts.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_opts.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off,semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_semijoin.result' --- a/mysql-test/r/subselect_no_semijoin.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_semijoin.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-15 19:52:58 +0000 @@ -202,39 +202,39 @@ BUG#37120 optimizer_switch allowable val select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 19:52:58 +0000 @@ -206,39 +206,39 @@ BUG#37120 optimizer_switch allowable val select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/t/ps.test' --- a/mysql-test/t/ps.test 2009-05-27 15:19:44 +0000 +++ b/mysql-test/t/ps.test 2010-03-11 21:43:31 +0000 @@ -163,6 +163,9 @@ create table t1 ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -171,6 +174,8 @@ explain SELECT (SELECT SUM(c1 + c12 + 0. deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # parameters from variables (for field creation) # === modified file 'mysql-test/t/subselect.test' --- a/mysql-test/t/subselect.test 2010-01-17 20:52:20 +0000 +++ b/mysql-test/t/subselect.test 2010-03-11 21:43:31 +0000 @@ -11,6 +11,9 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; --enable_warnings +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + select (select 2); explain extended select (select 2); SELECT (SELECT 1) UNION SELECT (SELECT 2); @@ -4061,4 +4064,6 @@ SELECT 1 FROM t1 GROUP BY (SELECT LAST_INSERT_ID() FROM t1 ORDER BY MIN(a) ASC LIMIT 1); DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; + --echo End of 5.1 tests. === modified file 'mysql-test/t/subselect3.test' --- a/mysql-test/t/subselect3.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect3.test 2010-03-11 21:43:31 +0000 @@ -59,9 +59,13 @@ select a in (select max(ie) from t1 wher show status like 'Handler_read_rnd_next'; select ' ^ This must show 11' Z; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + # This must show trigcond: explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; # @@ -529,6 +533,9 @@ SELECT a, MAX(b), DROP TABLE t1, t2; +# The next three test cases must be executed with the IN=>EXISTS strategy +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; # # Bug #27870: crash of an equijoin query with WHERE condition containing @@ -588,6 +595,8 @@ EXPLAIN SELECT a FROM t1 WHERE a NOT IN DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #34763: item_subselect.cc:1235:Item_in_subselect::row_value_transformer: # Assertion failed, unexpected error message: === modified file 'sql/item_cmpfunc.h' --- a/sql/item_cmpfunc.h 2010-03-13 20:04:52 +0000 +++ b/sql/item_cmpfunc.h 2010-03-15 19:52:58 +0000 @@ -350,6 +350,7 @@ public: CHARSET_INFO *compare_collation() { return cmp.cmp_collation.collation; } uint decimal_precision() const { return 1; } void top_level_item() { abort_on_null= TRUE; } + Arg_comparator *get_comparator() { return &cmp; } friend class Arg_comparator; }; === modified file 'sql/item_subselect.cc' --- a/sql/item_subselect.cc 2010-02-21 06:32:23 +0000 +++ b/sql/item_subselect.cc 2010-03-09 10:14:06 +0000 @@ -138,6 +138,7 @@ void Item_in_subselect::cleanup() left_expr_cache= NULL; } first_execution= TRUE; + is_constant= FALSE; Item_subselect::cleanup(); DBUG_VOID_RETURN; } @@ -449,8 +450,10 @@ bool Item_subselect::exec() int res; if (thd->is_error()) - /* Do not execute subselect in case of a fatal error */ + { + /* Do not execute subselect in case of a fatal error */ return 1; + } /* Simulate a failure in sub-query execution. Used to test e.g. out of memory or query being killed conditions. @@ -475,9 +478,6 @@ bool Item_subselect::exec() bool Item_in_subselect::exec() { DBUG_ENTER("Item_in_subselect::exec"); - DBUG_ASSERT(exec_method != MATERIALIZATION || - (exec_method == MATERIALIZATION && - engine->engine_type() == subselect_engine::HASH_SJ_ENGINE)); /* Initialize the cache of the left predicate operand. This has to be done as late as now, because Cached_item directly contains a resolved field (not @@ -493,14 +493,14 @@ bool Item_in_subselect::exec() if (!left_expr_cache && exec_method == MATERIALIZATION) init_left_expr_cache(); - /* If the new left operand is already in the cache, reuse the old result. */ - if (left_expr_cache && test_if_item_cache_changed(*left_expr_cache) < 0) - { - /* Always compute IN for the first row as the cache is not valid for it. */ - if (!first_execution) - DBUG_RETURN(FALSE); - first_execution= FALSE; - } + /* + If the new left operand is already in the cache, reuse the old result. + Use the cached result only if this is not the first execution of IN + because the cache is not valid for the first execution. + */ + if (!first_execution && left_expr_cache && + test_if_item_cache_changed(*left_expr_cache) < 0) + DBUG_RETURN(FALSE); /* The exec() method below updates item::value, and item::null_value, thus if @@ -910,8 +910,8 @@ bool Item_in_subselect::test_limit(st_se Item_in_subselect::Item_in_subselect(Item * left_exp, st_select_lex *select_lex): Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), - upper_item(0) + is_constant(FALSE), optimizer(0), pushed_cond_guards(NULL), + exec_method(NOT_TRANSFORMED), upper_item(0) { DBUG_ENTER("Item_in_subselect::Item_in_subselect"); left_expr= left_exp; @@ -1105,6 +1105,8 @@ bool Item_in_subselect::val_bool() { DBUG_ASSERT(fixed == 1); null_value= 0; + if (is_constant) + return value; if (exec()) { reset(); @@ -1571,9 +1573,9 @@ Item_in_subselect::row_value_transformer DBUG_ENTER("Item_in_subselect::row_value_transformer"); // psergey: duplicated_subselect_card_check - if (select_lex->item_list.elements != left_expr->cols()) + if (select_lex->item_list.elements != cols_num) { - my_error(ER_OPERAND_COLUMNS, MYF(0), left_expr->cols()); + my_error(ER_OPERAND_COLUMNS, MYF(0), cols_num); DBUG_RETURN(RES_ERROR); } @@ -1980,17 +1982,69 @@ void Item_in_subselect::print(String *st bool Item_in_subselect::fix_fields(THD *thd_arg, Item **ref) { - bool result = 0; + uint outer_cols_num; + List<Item> *inner_cols; if (exec_method == SEMI_JOIN) return !( (*ref)= new Item_int(1)); - if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed) - result = left_expr->fix_fields(thd_arg, &left_expr); + /* + Check if the outer and inner IN operands match in those cases when we + will not perform IN=>EXISTS transformation. Currently this is when we + use subquery materialization. + + The condition below is true when this method was called recursively from + inside JOIN::prepare for the JOIN object created by the call chain + Item_subselect::fix_fields -> subselect_single_select_engine::prepare, + which creates a JOIN object for the subquery and calls JOIN::prepare for + the JOIN of the subquery. + Notice that in some cases, this doesn't happen, and the check_cols() + test for each Item happens later in + Item_in_subselect::row_value_in_to_exists_transformer. + The reason for this mess is that our JOIN::prepare phase works top-down + instead of bottom-up, so we first do name resoluton and semantic checks + for the outer selects, then for the inner. + */ + if (engine && + engine->engine_type() == subselect_engine::SINGLE_SELECT_ENGINE && + ((subselect_single_select_engine*)engine)->join) + { + outer_cols_num= left_expr->cols(); + + if (unit->is_union()) + inner_cols= &(unit->types); + else + inner_cols= &(unit->first_select()->item_list); + if (outer_cols_num != inner_cols->elements) + { + my_error(ER_OPERAND_COLUMNS, MYF(0), outer_cols_num); + return TRUE; + } + if (outer_cols_num > 1) + { + List_iterator<Item> inner_col_it(*inner_cols); + Item *inner_col; + for (uint i= 0; i < outer_cols_num; i++) + { + inner_col= inner_col_it++; + if (inner_col->check_cols(left_expr->element_index(i)->cols())) + return TRUE; + } + } + } + + if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed && + left_expr->fix_fields(thd_arg, &left_expr)) + return TRUE; + if (Item_subselect::fix_fields(thd_arg, ref)) + return TRUE; - return result || Item_subselect::fix_fields(thd_arg, ref); + fixed= TRUE; + + return FALSE; } + void Item_in_subselect::fix_after_pullout(st_select_lex *new_parent, Item **ref) { left_expr->fix_after_pullout(new_parent, &left_expr); @@ -2267,10 +2321,9 @@ bool subselect_union_engine::no_rows() void subselect_uniquesubquery_engine::cleanup() { DBUG_ENTER("subselect_uniquesubquery_engine::cleanup"); - /* - subselect_uniquesubquery_engine have not 'result' assigbed, so we do not - cleanup() it - */ + /* Tell handler we don't need the index anymore */ + if (tab->table->file->inited) + tab->table->file->ha_index_end(); DBUG_VOID_RETURN; } @@ -2291,7 +2344,7 @@ subselect_union_engine::subselect_union_ Create and prepare the JOIN object that represents the query execution plan for the subquery. - @detail + @details This method is called from Item_subselect::fix_fields. For prepared statements it is called both during the PREPARE and EXECUTE phases in the following ways: @@ -2593,14 +2646,23 @@ int subselect_uniquesubquery_engine::sca for (;;) { error=table->file->ha_rnd_next(table->record[0]); - if (error && error != HA_ERR_END_OF_FILE) - { - error= report_error(table, error); - break; + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(table, error); + break; + } } - /* No more rows */ - if (table->status) - break; if (!cond || cond->val_int()) { @@ -2711,6 +2773,56 @@ bool subselect_uniquesubquery_engine::co /* + @retval 1 A NULL was found in the outer reference, index lookup is + not applicable, the outer ref is unsusable as a lookup key, + use some other method to find a match. + @retval 0 The outer ref was copied into an index lookup key. + @retval -1 The outer ref cannot possibly match any row, IN is FALSE. +*/ +/* TIMOUR: this method is a variant of copy_ref_key(), needs refactoring. */ + +int subselect_uniquesubquery_engine::copy_ref_key_simple() +{ + for (store_key **copy= tab->ref.key_copy ; *copy ; copy++) + { + enum store_key::store_key_result store_res; + store_res= (*copy)->copy(); + tab->ref.key_err= store_res; + + /* + When there is a NULL part in the key we don't need to make index + lookup for such key thus we don't need to copy whole key. + If we later should do a sequential scan return OK. Fail otherwise. + + See also the comment for the subselect_uniquesubquery_engine::exec() + function. + */ + null_keypart= (*copy)->null_key; + if (null_keypart) + return 1; + + /* + Check if the error is equal to STORE_KEY_FATAL. This is not expressed + using the store_key::store_key_result enum because ref.key_err is a + boolean and we want to detect both TRUE and STORE_KEY_FATAL from the + space of the union of the values of [TRUE, FALSE] and + store_key::store_key_result. + TODO: fix the variable an return types. + */ + if (store_res == store_key::STORE_KEY_FATAL) + { + /* + Error converting the left IN operand to the column type of the right + IN operand. + */ + return -1; + } + } + return 0; +} + + +/* Execute subselect SYNOPSIS @@ -2750,7 +2862,13 @@ int subselect_uniquesubquery_engine::exe /* TODO: change to use of 'full_scan' here? */ if (copy_ref_key()) + { + /* + TIMOUR: copy_ref_key() == 1 means NULL result, not error, why return 1? + Check who reiles on this result. + */ DBUG_RETURN(1); + } if (table->status) { /* @@ -2791,6 +2909,46 @@ int subselect_uniquesubquery_engine::exe } +/* + TIMOUR: write comment +*/ + +int subselect_uniquesubquery_engine::index_lookup() +{ + DBUG_ENTER("subselect_uniquesubquery_engine::index_lookup"); + int error; + TABLE *table= tab->table; + + if (!table->file->inited) + table->file->ha_index_init(tab->ref.key, 0); + error= table->file->ha_index_read_map(table->record[0], + tab->ref.key_buff, + make_prev_keypart_map(tab-> + ref.key_parts), + HA_READ_KEY_EXACT); + DBUG_PRINT("info", ("lookup result: %i", error)); + + if (error && error != HA_ERR_KEY_NOT_FOUND && error != HA_ERR_END_OF_FILE) + { + /* + TIMOUR: I don't understand at all when do we need to call report_error. + In most places where we access an index, we don't do this. Why here? + */ + error= report_error(table, error); + DBUG_RETURN(error); + } + + table->null_row= 0; + if (!error && (!cond || cond->val_int())) + ((Item_in_subselect *) item)->value= 1; + else + ((Item_in_subselect *) item)->value= 0; + + DBUG_RETURN(0); +} + + + subselect_uniquesubquery_engine::~subselect_uniquesubquery_engine() { /* Tell handler we don't need the index anymore */ @@ -3225,6 +3383,7 @@ bool subselect_union_engine::no_tables() bool subselect_uniquesubquery_engine::no_tables() { /* returning value is correct, but this method should never be called */ + DBUG_ASSERT(FALSE); return 0; } @@ -3235,16 +3394,259 @@ bool subselect_uniquesubquery_engine::no /** + Check if an IN predicate should be executed via partial matching using + only schema information. + + @details + This test essentially has three results: + - partial matching is applicable, but cannot be executed due to a + limitation in the total number of indexes, as a result we can't + use subquery materialization at all. + - partial matching is either applicable or not, and this can be + determined by looking at 'this->max_keys'. + If max_keys > 1, then we need partial matching because there are + more indexes than just the one we use during materialization to + remove duplicates. + + @note + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_schema() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + + if (item_in->is_top_level_item()) + return COMPLETE_MATCH; + else + { + List_iterator<Item> inner_col_it(*item_in->unit->get_unit_column_types()); + Item *outer_col, *inner_col; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + outer_col= item_in->left_expr->element_index(i); + inner_col= inner_col_it++; + + if (!inner_col->maybe_null && !outer_col->maybe_null) + bitmap_set_bit(&non_null_key_parts, i); + else + { + bitmap_set_bit(&partial_match_key_parts, i); + ++count_partial_match_columns; + } + } + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (count_partial_match_columns) + return PARTIAL_MATCH; + return COMPLETE_MATCH; +} + + +/** + Test whether an IN predicate must be computed via partial matching + based on the NULL statistics for each column of a materialized subquery. + + @details The procedure analyzes column NULL statistics, updates the + matching type of columns that cannot be NULL or that contain only NULLs. + Based on this, the procedure determines the final execution strategy for + the [NOT] IN predicate. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_data() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + Item *outer_col; + + /* + If we already determined that a complete match is enough based on schema + information, nothing can be better. + */ + if (strategy == COMPLETE_MATCH) + return COMPLETE_MATCH; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + if (!bitmap_is_set(&partial_match_key_parts, i)) + continue; + outer_col= item_in->left_expr->element_index(i); + /* + If column 'i' doesn't contain NULLs, and the corresponding outer reference + cannot have a NULL value, then 'i' is a non-nullable column. + */ + if (result_sink->get_null_count_of_col(i) == 0 && !outer_col->maybe_null) + { + bitmap_clear_bit(&partial_match_key_parts, i); + bitmap_set_bit(&non_null_key_parts, i); + --count_partial_match_columns; + } + if (result_sink->get_null_count_of_col(i) == + tmp_table->file->stats.records) + ++count_null_only_columns; + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (!count_partial_match_columns) + return COMPLETE_MATCH; + return PARTIAL_MATCH; +} + + +void +subselect_hash_sj_engine::choose_partial_match_strategy( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t pm_buff_size; + + DBUG_ASSERT(strategy == PARTIAL_MATCH); + /* + Choose according to global optimizer switch. If only one of the switches is + 'ON', then the remaining strategy is the only possible one. The only cases + when this will be overriden is when the total size of all buffers for the + merge strategy is bigger than the 'rowid_merge_buff_size' system variable, + or if there isn't enough physical memory to allocate the buffers. + */ + if (!optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_SCAN; + else if + ( optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + !optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_MERGE; + + /* + If both switches are ON, or both are OFF, we interpret that as "let the + optimizer decide". Perform a cost based choice between the two partial + matching strategies. + */ + /* + TIMOUR: the above interpretation of the switch values could be changed to: + - if both are ON - let the optimizer decide, + - if both are OFF - do not use partial matching, therefore do not use + materialization in non-top-level predicates. + The problem with this is that we know for sure if we need partial matching + only after the subquery is materialized, and this is too late to revert to + the IN=>EXISTS strategy. + */ + if (strategy == PARTIAL_MATCH) + { + /* + TIMOUR: Currently we use a super simplistic measure. This will be + addressed in a separate task. + */ + if (tmp_table->file->stats.records < 100) + strategy= PARTIAL_MATCH_SCAN; + else + strategy= PARTIAL_MATCH_MERGE; + } + + /* Check if there is enough memory for the rowid merge strategy. */ + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_buff_size= rowid_merge_buff_size(has_non_null_key, + has_covering_null_row, + partial_match_key_parts); + if (pm_buff_size > thd->variables.rowid_merge_buff_size) + strategy= PARTIAL_MATCH_SCAN; + } +} + + +/* + Compute the memory size of all buffers proportional to the number of rows + in tmp_table. + + @details + If the result is bigger than thd->variables.rowid_merge_buff_size, partial + matching via merging is not applicable. +*/ + +size_t subselect_hash_sj_engine::rowid_merge_buff_size( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t buff_size; /* Total size of all buffers used by partial matching. */ + ha_rows row_count= tmp_table->file->stats.records; + uint rowid_length= tmp_table->file->ref_length; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + /* Size of the subselect_rowid_merge_engine::row_num_to_rowid buffer. */ + buff_size= row_count * rowid_length * sizeof(uchar); + + if (has_non_null_key) + { + /* Add the size of Ordered_key::key_buff of the only non-NULL key. */ + buff_size+= row_count * sizeof(rownum_t); + } + + if (!has_covering_null_row) + { + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i) || + result_sink->get_null_count_of_col(i) == row_count) + continue; /* In these cases we wouldn't construct Ordered keys. */ + + /* Add the size of Ordered_key::key_buff */ + buff_size+= (row_count - result_sink->get_null_count_of_col(i)) * + sizeof(rownum_t); + /* Add the size of Ordered_key::null_key */ + buff_size+= bitmap_buffer_size(result_sink->get_max_null_of_col(i)); + } + } + + return buff_size; +} + + +/* + Initialize a MY_BITMAP with a buffer allocated on the current + memory root. + TIMOUR: move to bitmap C file? +*/ + +static my_bool +bitmap_init_memroot(MY_BITMAP *map, uint n_bits, MEM_ROOT *mem_root) +{ + my_bitmap_map *bitmap_buf; + + if (!(bitmap_buf= (my_bitmap_map*) alloc_root(mem_root, + bitmap_buffer_size(n_bits))) || + bitmap_init(map, bitmap_buf, n_bits, FALSE)) + return TRUE; + bitmap_clear_all(map); + return FALSE; +} + + +/** Create all structures needed for IN execution that can live between PS reexecution. - @detail + @param tmp_columns the items that produce the data for the temp table + + @details - Create a temporary table to store the result of the IN subquery. The temporary table has one hash index on all its columns. - Create a new result sink that sends the result stream of the subquery to the temporary table, - - Create and initialize a new JOIN_TAB, and TABLE_REF objects to perform - lookups into the indexed temporary table. @notice: Currently Item_subselect::init() already chooses and creates at parse @@ -3256,145 +3658,210 @@ bool subselect_uniquesubquery_engine::no bool subselect_hash_sj_engine::init_permanent(List<Item> *tmp_columns) { - /* The result sink where we will materialize the subquery result. */ - select_union *tmp_result_sink; - /* The table into which the subquery is materialized. */ - TABLE *tmp_table; - KEY *tmp_key; /* The only index on the temporary table. */ - uint tmp_key_parts; /* Number of keyparts in tmp_key. */ - Item_in_subselect *item_in= (Item_in_subselect *) item; + /* Options to create_tmp_table. */ + ulonglong tmp_create_options= thd->options | TMP_TABLE_ALL_COLUMNS; + /* | TMP_TABLE_FORCE_MYISAM; TIMOUR: force MYISAM */ DBUG_ENTER("subselect_hash_sj_engine::init_permanent"); - /* 1. Create/initialize materialization related objects. */ + if (bitmap_init_memroot(&non_null_key_parts, tmp_columns->elements, + thd->mem_root) || + bitmap_init_memroot(&partial_match_key_parts, tmp_columns->elements, + thd->mem_root)) + DBUG_RETURN(TRUE); /* Create and initialize a select result interceptor that stores the result stream in a temporary table. The temporary table itself is managed (created/filled/etc) internally by the interceptor. */ - if (!(tmp_result_sink= new select_union)) +/* + TIMOUR: + Select a more efficient result sink when we know there is no need to collect + data statistics. + + if (strategy == COMPLETE_MATCH) + { + if (!(result= new select_union)) + DBUG_RETURN(TRUE); + } + else if (strategy == PARTIAL_MATCH) + { + if (!(result= new select_materialize_with_stats)) + DBUG_RETURN(TRUE); + } +*/ + if (!(result= new select_materialize_with_stats)) DBUG_RETURN(TRUE); - if (tmp_result_sink->create_result_table( - thd, tmp_columns, TRUE, - thd->options | TMP_TABLE_ALL_COLUMNS, + + if (((select_union*) result)->create_result_table( + thd, tmp_columns, TRUE, tmp_create_options, "materialized subselect", TRUE)) DBUG_RETURN(TRUE); - tmp_table= tmp_result_sink->table; - tmp_key= tmp_table->key_info; - tmp_key_parts= tmp_key->key_parts; + tmp_table= ((select_union*) result)->table; /* - If the subquery has blobs, or the total key lenght is bigger than some - length, then the created index cannot be used for lookups and we - can't use hash semi join. If this is the case, delete the temporary - table since it will not be used, and tell the caller we failed to - initialize the engine. + If the subquery has blobs, or the total key lenght is bigger than + some length, or the total number of key parts is more than the + allowed maximum (currently MAX_REF_PARTS == 16), then the created + index cannot be used for lookups and we can't use hash semi + join. If this is the case, delete the temporary table since it + will not be used, and tell the caller we failed to initialize the + engine. */ if (tmp_table->s->keys == 0) { -#ifndef DBUG_OFF - handlerton *tmp_table_hton= tmp_table->s->db_type(); -#ifdef USE_MARIA_FOR_TMP_TABLES - DBUG_ASSERT(tmp_table_hton == maria_hton); -#else - DBUG_ASSERT(tmp_table_hton == myisam_hton); -#endif -#endif DBUG_ASSERT( tmp_table->s->uniques || tmp_table->key_info->key_length >= tmp_table->file->max_key_length() || tmp_table->key_info->key_parts > tmp_table->file->max_key_parts()); free_tmp_table(thd, tmp_table); + tmp_table= NULL; delete result; result= NULL; DBUG_RETURN(TRUE); } - result= tmp_result_sink; /* Make sure there is only one index on the temp table, and it doesn't have the extra key part created when s->uniques > 0. */ - DBUG_ASSERT(tmp_table->s->keys == 1 && tmp_columns->elements == tmp_key_parts); + DBUG_ASSERT(tmp_table->s->keys == 1 && + ((Item_in_subselect *) item)->left_expr->cols() == + tmp_table->key_info->key_parts); + + if (make_semi_join_conds() || + /* A unique_engine is used both for complete and partial matching. */ + !(lookup_engine= make_unique_engine())) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} - /* 2. Create/initialize execution related objects. */ +/* + Create an artificial condition to post-filter those rows matched by index + lookups that cannot be distinguished by the index lookup procedure. - /* - Create and initialize the JOIN_TAB that represents an index lookup - plan operator into the materialized subquery result. Notice that: - - this JOIN_TAB has no corresponding JOIN (and doesn't need one), and - - here we initialize only those members that are used by - subselect_uniquesubquery_engine, so these objects are incomplete. - */ - if (!(tab= (JOIN_TAB*) thd->alloc(sizeof(JOIN_TAB)))) - DBUG_RETURN(TRUE); - tab->table= tmp_table; - tab->ref.key= 0; /* The only temp table index. */ - tab->ref.key_length= tmp_key->key_length; - if (!(tab->ref.key_buff= - (uchar*) thd->calloc(ALIGN_SIZE(tmp_key->key_length) * 2)) || - !(tab->ref.key_copy= - (store_key**) thd->alloc((sizeof(store_key*) * - (tmp_key_parts + 1)))) || - !(tab->ref.items= - (Item**) thd->alloc(sizeof(Item*) * tmp_key_parts))) - DBUG_RETURN(TRUE); + @notes + The need for post-filtering may occur e.g. because of + truncation. Prepared statements execution requires that fix_fields is + called for every execution. In order to call fix_fields we need to + create a Name_resolution_context and a corresponding TABLE_LIST for + the temporary table for the subquery, so that all column references + to the materialized subquery table can be resolved correctly. - KEY_PART_INFO *cur_key_part= tmp_key->key_part; - store_key **ref_key= tab->ref.key_copy; - uchar *cur_ref_buff= tab->ref.key_buff; + @returns + @retval TRUE memory allocation error occurred + @retval FALSE the conditions were created and resolved (fixed) +*/ - /* - Create an artificial condition to post-filter those rows matched by index - lookups that cannot be distinguished by the index lookup procedure, e.g. - because of truncation. Prepared statements execution requires that - fix_fields is called for every execution. In order to call fix_fields we - need to create a Name_resolution_context and a corresponding TABLE_LIST - for the temporary table for the subquery, so that all column references - to the materialized subquery table can be resolved correctly. - */ - DBUG_ASSERT(cond == NULL); - if (!(cond= new Item_cond_and)) - DBUG_RETURN(TRUE); +bool subselect_hash_sj_engine::make_semi_join_conds() +{ /* Table reference for tmp_table that is used to resolve column references (Item_fields) to columns in tmp_table. */ TABLE_LIST *tmp_table_ref; + /* Name resolution context for all tmp_table columns created below. */ + Name_resolution_context *context; + Item_in_subselect *item_in= (Item_in_subselect *) item; + + DBUG_ENTER("subselect_hash_sj_engine::make_semi_join_conds"); + DBUG_ASSERT(semi_join_conds == NULL); + + if (!(semi_join_conds= new Item_cond_and)) + DBUG_RETURN(TRUE); + if (!(tmp_table_ref= (TABLE_LIST*) thd->alloc(sizeof(TABLE_LIST)))) DBUG_RETURN(TRUE); tmp_table_ref->init_one_table("", "materialized subselect", TL_READ); tmp_table_ref->table= tmp_table; - /* Name resolution context for all tmp_table columns created below. */ - Name_resolution_context *context= new Name_resolution_context; + context= new Name_resolution_context; context->init(); context->first_name_resolution_table= context->last_name_resolution_table= tmp_table_ref; - for (uint i= 0; i < tmp_key_parts; i++, cur_key_part++, ref_key++) + for (uint i= 0; i < item_in->left_expr->cols(); i++) { Item_func_eq *eq_cond; /* New equi-join condition for the current column. */ /* Item for the corresponding field from the materialized temp table. */ Item_field *right_col_item; - int null_count= test(cur_key_part->field->real_maybe_null()); - tab->ref.items[i]= item_in->left_expr->element_index(i); - if (!(right_col_item= new Item_field(thd, context, cur_key_part->field)) || - !(eq_cond= new Item_func_eq(tab->ref.items[i], right_col_item)) || - ((Item_cond_and*)cond)->add(eq_cond)) + if (!(right_col_item= new Item_field(thd, context, tmp_table->field[i])) || + !(eq_cond= new Item_func_eq(item_in->left_expr->element_index(i), + right_col_item)) || + (((Item_cond_and*)semi_join_conds)->add(eq_cond))) { - delete cond; - cond= NULL; + delete semi_join_conds; + semi_join_conds= NULL; DBUG_RETURN(TRUE); } + } + if (semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} + + +/** + Create a new uniquesubquery engine for the execution of an IN predicate. + + @details + Create and initialize a new JOIN_TAB, and Table_ref objects to perform + lookups into the indexed temporary table. + + @retval A new subselect_hash_sj_engine object + @retval NULL if a memory allocation error occurs +*/ + +subselect_uniquesubquery_engine* +subselect_hash_sj_engine::make_unique_engine() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + /* The only index on the temporary table. */ + KEY *tmp_key= tmp_table->key_info; + /* Number of keyparts in tmp_key. */ + uint tmp_key_parts= tmp_key->key_parts; + JOIN_TAB *tab; + + DBUG_ENTER("subselect_hash_sj_engine::make_unique_engine"); + + /* + Create and initialize the JOIN_TAB that represents an index lookup + plan operator into the materialized subquery result. Notice that: + - this JOIN_TAB has no corresponding JOIN (and doesn't need one), and + - here we initialize only those members that are used by + subselect_uniquesubquery_engine, so these objects are incomplete. + */ + if (!(tab= (JOIN_TAB*) thd->alloc(sizeof(JOIN_TAB)))) + DBUG_RETURN(NULL); + tab->table= tmp_table; + tab->ref.key= 0; /* The only temp table index. */ + tab->ref.key_length= tmp_key->key_length; + if (!(tab->ref.key_buff= + (uchar*) thd->calloc(ALIGN_SIZE(tmp_key->key_length) * 2)) || + !(tab->ref.key_copy= + (store_key**) thd->alloc((sizeof(store_key*) * + (tmp_key_parts + 1)))) || + !(tab->ref.items= + (Item**) thd->alloc(sizeof(Item*) * tmp_key_parts))) + DBUG_RETURN(NULL); + KEY_PART_INFO *cur_key_part= tmp_key->key_part; + store_key **ref_key= tab->ref.key_copy; + uchar *cur_ref_buff= tab->ref.key_buff; + + for (uint i= 0; i < tmp_key_parts; i++, cur_key_part++, ref_key++) + { + tab->ref.items[i]= item_in->left_expr->element_index(i); + int null_count= test(cur_key_part->field->real_maybe_null()); *ref_key= new store_key_item(thd, cur_key_part->field, - /* TODO: + /* TIMOUR: the NULL byte is taken into account in cur_key_part->store_length, so instead of cur_ref_buff + test(maybe_null), we could @@ -3409,10 +3876,8 @@ bool subselect_hash_sj_engine::init_perm tab->ref.key_err= 1; tab->ref.key_parts= tmp_key_parts; - if (cond->fix_fields(thd, &cond)) - DBUG_RETURN(TRUE); - - DBUG_RETURN(FALSE); + DBUG_RETURN(new subselect_uniquesubquery_engine(thd, tab, item, + semi_join_conds)); } @@ -3435,7 +3900,8 @@ bool subselect_hash_sj_engine::init_runt Repeat name resolution for 'cond' since cond is not part of any clause of the query, and it is not 'fixed' during JOIN::prepare. */ - if (cond && !cond->fixed && cond->fix_fields(thd, &cond)) + if (semi_join_conds && !semi_join_conds->fixed && + semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) return TRUE; /* Let our engine reuse this query plan for materialization. */ materialize_join= materialize_engine->join; @@ -3446,32 +3912,53 @@ bool subselect_hash_sj_engine::init_runt subselect_hash_sj_engine::~subselect_hash_sj_engine() { + delete lookup_engine; delete result; - if (tab) - free_tmp_table(thd, tab->table); + if (tmp_table) + free_tmp_table(thd, tmp_table); } /** Cleanup performed after each PS execution. - @detail + @details Called in the end of JOIN::prepare for PS from Item_subselect::cleanup. */ void subselect_hash_sj_engine::cleanup() { + enum_engine_type lookup_engine_type= lookup_engine->engine_type(); is_materialized= FALSE; - result->cleanup(); /* Resets the temp table as well. */ + bitmap_clear_all(&non_null_key_parts); + bitmap_clear_all(&partial_match_key_parts); + count_partial_match_columns= 0; + count_null_only_columns= 0; + strategy= UNDEFINED; materialize_engine->cleanup(); - subselect_uniquesubquery_engine::cleanup(); + if (lookup_engine_type == TABLE_SCAN_ENGINE || + lookup_engine_type == ROWID_MERGE_ENGINE) + { + subselect_engine *inner_lookup_engine; + inner_lookup_engine= + ((subselect_partial_match_engine*) lookup_engine)->lookup_engine; + /* + Partial match engines are recreated for each PS execution inside + subselect_hash_sj_engine::exec(). + */ + delete lookup_engine; + lookup_engine= inner_lookup_engine; + } + DBUG_ASSERT(lookup_engine->engine_type() == UNIQUESUBQUERY_ENGINE); + lookup_engine->cleanup(); + result->cleanup(); /* Resets the temp table as well. */ } /** Execute a subquery IN predicate via materialization. - @detail + @details If needed materialize the subquery into a temporary table, then copmpute the predicate via a lookup into this table. @@ -3482,6 +3969,9 @@ void subselect_hash_sj_engine::cleanup() int subselect_hash_sj_engine::exec() { Item_in_subselect *item_in= (Item_in_subselect *) item; + SELECT_LEX *save_select= thd->lex->current_select; + subselect_partial_match_engine *pm_engine= NULL; + int res= 0; DBUG_ENTER("subselect_hash_sj_engine::exec"); @@ -3489,56 +3979,126 @@ int subselect_hash_sj_engine::exec() Optimize and materialize the subquery during the first execution of the subquery predicate. */ - if (!is_materialized) - { - int res= 0; - SELECT_LEX *save_select= thd->lex->current_select; - thd->lex->current_select= materialize_engine->select_lex; - if ((res= materialize_join->optimize())) - goto err; /* purecov: inspected */ - materialize_join->exec(); - if ((res= test(materialize_join->error || thd->is_fatal_error))) - goto err; - - /* - TODO: - - Unlock all subquery tables as we don't need them. To implement this - we need to add new functionality to JOIN::join_free that can unlock - all tables in a subquery (and all its subqueries). - - The temp table used for grouping in the subquery can be freed - immediately after materialization (yet it's done together with - unlocking). - */ - is_materialized= TRUE; - /* - If the subquery returned no rows, the temporary table is empty, so we know - directly that the result of IN is FALSE. We first update the table - statistics, then we test if the temporary table for the query result is - empty. - */ - tab->table->file->info(HA_STATUS_VARIABLE); - if (!tab->table->file->stats.records) - { - empty_result_set= TRUE; - item_in->value= FALSE; - /* TODO: check we need this: item_in->null_value= FALSE; */ - DBUG_RETURN(FALSE); - } - /* Set tmp_param only if its usable, i.e. tmp_param->copy_field != NULL. */ - tmp_param= &(item_in->unit->outer_select()->join->tmp_table_param); - if (tmp_param && !tmp_param->copy_field) - tmp_param= NULL; + thd->lex->current_select= materialize_engine->select_lex; + if ((res= materialize_join->optimize())) + goto err; /* purecov: inspected */ + DBUG_ASSERT(!is_materialized); /* We should materialize only once. */ + materialize_join->exec(); + if ((res= test(materialize_join->error || thd->is_fatal_error))) + goto err; -err: - thd->lex->current_select= save_select; - if (res) - DBUG_RETURN(res); + /* + TODO: + - Unlock all subquery tables as we don't need them. To implement this + we need to add new functionality to JOIN::join_free that can unlock + all tables in a subquery (and all its subqueries). + - The temp table used for grouping in the subquery can be freed + immediately after materialization (yet it's done together with + unlocking). + */ + is_materialized= TRUE; + /* + If the subquery returned no rows, the temporary table is empty, so we know + directly that the result of IN is FALSE. We first update the table + statistics, then we test if the temporary table for the query result is + empty. + */ + tmp_table->file->info(HA_STATUS_VARIABLE); + if (!tmp_table->file->stats.records) + { + item_in->value= FALSE; + /* The value of IN will not change during this execution. */ + item_in->is_constant= TRUE; + item_in->set_first_execution(); + /* TIMOUR: check if we need this: item_in->null_value= FALSE; */ + DBUG_RETURN(FALSE); } /* - Lookup the left IN operand in the hash index of the materialized subquery. + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. */ - DBUG_RETURN(subselect_uniquesubquery_engine::exec()); + strategy= get_strategy_using_schema(); + /* This call may discover that we don't need partial matching at all. */ + strategy= get_strategy_using_data(); + if (strategy == PARTIAL_MATCH) + { + uint count_pm_keys; /* Total number of keys needed for partial matching. */ + MY_BITMAP *nn_key_parts; /* The key parts of the only non-NULL index. */ + uint covering_null_row_width; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + nn_key_parts= (count_partial_match_columns < tmp_table->s->fields) ? + &non_null_key_parts : NULL; + + if (result_sink->get_max_nulls_in_row() == + tmp_table->s->fields - + (nn_key_parts ? bitmap_bits_set(nn_key_parts) : 0)) + covering_null_row_width= result_sink->get_max_nulls_in_row(); + else + covering_null_row_width= 0; + + if (covering_null_row_width) + count_pm_keys= nn_key_parts ? 1 : 0; + else + count_pm_keys= count_partial_match_columns - count_null_only_columns + + (nn_key_parts ? 1 : 0); + + choose_partial_match_strategy(test(nn_key_parts), + test(covering_null_row_width), + &partial_match_key_parts); + DBUG_ASSERT(strategy == PARTIAL_MATCH_MERGE || + strategy == PARTIAL_MATCH_SCAN); + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_engine= + new subselect_rowid_merge_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + count_pm_keys, + covering_null_row_width, + item, result, + semi_join_conds->argument_list()); + if (!pm_engine || + ((subselect_rowid_merge_engine*) pm_engine)-> + init(nn_key_parts, &partial_match_key_parts)) + { + /* + The call to init() would fail if there was not enough memory to allocate + all buffers for the rowid merge strategy. In this case revert to table + scanning which doesn't need any big buffers. + */ + delete pm_engine; + pm_engine= NULL; + strategy= PARTIAL_MATCH_SCAN; + } + } + + if (strategy == PARTIAL_MATCH_SCAN) + { + if (!(pm_engine= + new subselect_table_scan_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + item, result, + semi_join_conds->argument_list(), + covering_null_row_width))) + { + /* This is an irrecoverable error. */ + res= 1; + goto err; + } + } + } + + if (pm_engine) + lookup_engine= pm_engine; + item_in->change_engine(lookup_engine); + +err: + thd->lex->current_select= save_select; + DBUG_RETURN(res); } @@ -3551,10 +4111,1008 @@ void subselect_hash_sj_engine::print(Str str->append(STRING_WITH_LEN(" <materialize> (")); materialize_engine->print(str, query_type); str->append(STRING_WITH_LEN(" ), ")); - if (tab) - subselect_uniquesubquery_engine::print(str, query_type); + + if (lookup_engine) + lookup_engine->print(str, query_type); else str->append(STRING_WITH_LEN( - "<the access method for lookups is not yet created>" + "<engine selected at execution time>" )); } + +void subselect_hash_sj_engine::fix_length_and_dec(Item_cache** row) +{ + DBUG_ASSERT(FALSE); +} + +void subselect_hash_sj_engine::exclude() +{ + DBUG_ASSERT(FALSE); +} + +bool subselect_hash_sj_engine::no_tables() +{ + DBUG_ASSERT(FALSE); + return FALSE; +} + +bool subselect_hash_sj_engine::change_result(Item_subselect *si, + select_result_interceptor *res) +{ + DBUG_ASSERT(FALSE); + return TRUE; +} + + +Ordered_key::Ordered_key(uint keyid_arg, TABLE *tbl_arg, Item *search_key_arg, + ha_rows null_count_arg, ha_rows min_null_row_arg, + ha_rows max_null_row_arg, uchar *row_num_to_rowid_arg) + : keyid(keyid_arg), tbl(tbl_arg), search_key(search_key_arg), + row_num_to_rowid(row_num_to_rowid_arg), null_count(null_count_arg) +{ + DBUG_ASSERT(tbl->file->stats.records > null_count); + key_buff_elements= tbl->file->stats.records - null_count; + cur_key_idx= HA_POS_ERROR; + + DBUG_ASSERT((null_count && min_null_row_arg && max_null_row_arg) || + (!null_count && !min_null_row_arg && !max_null_row_arg)); + if (null_count) + { + /* The counters are 1-based, for key access we need 0-based indexes. */ + min_null_row= min_null_row_arg - 1; + max_null_row= max_null_row_arg - 1; + } + else + min_null_row= max_null_row= 0; +} + + +Ordered_key::~Ordered_key() +{ + my_free((char*) key_buff, MYF(0)); + bitmap_free(&null_key); +} + + +/* + Cleanup that needs to be done for each PS (re)execution. +*/ + +void Ordered_key::cleanup() +{ + /* + Currently these keys are recreated for each PS re-execution, thus + there is nothing to cleanup, the whole object goes away after execution + is over. All handler related initialization/deinitialization is done by + the parent subselect_rowid_merge_engine object. + */ +} + + +/* + Initialize a multi-column index. +*/ + +bool Ordered_key::init(MY_BITMAP *columns_to_index) +{ + THD *thd= tbl->in_use; + uint cur_key_col= 0; + Item_field *cur_tmp_field; + Item_func_lt *fn_less_than; + + key_column_count= bitmap_bits_set(columns_to_index); + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(key_column_count * + sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(key_column_count * + sizeof(Item_func_lt*)); + + for (uint i= 0; i < columns_to_index->n_bits; i++) + { + if (!bitmap_is_set(columns_to_index, i)) + continue; + cur_tmp_field= new Item_field(tbl->field[i]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + fn_less_than= new Item_func_lt(cur_tmp_field, + search_key->element_index(i)); + fn_less_than->fix_fields(thd, (Item**) &fn_less_than); + key_columns[cur_key_col]= cur_tmp_field; + compare_pred[cur_key_col]= fn_less_than; + ++cur_key_col; + } + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Initialize a single-column index. +*/ + +bool Ordered_key::init(int col_idx) +{ + THD *thd= tbl->in_use; + + key_column_count= 1; + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(sizeof(Item_func_lt*)); + + key_columns[0]= new Item_field(tbl->field[col_idx]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + compare_pred[0]= new Item_func_lt(key_columns[0], + search_key->element_index(col_idx)); + compare_pred[0]->fix_fields(thd, (Item**)&compare_pred[0]); + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Allocate the buffers for both the row number, and the NULL-bitmap indexes. +*/ + +bool Ordered_key::alloc_keys_buffers() +{ + DBUG_ASSERT(key_buff_elements > 0); + + if (!(key_buff= (rownum_t*) my_malloc(key_buff_elements * sizeof(rownum_t), + MYF(MY_WME)))) + return TRUE; + + /* + TIMOUR: it is enough to create bitmaps with size + (max_null_row - min_null_row), and then use min_null_row as + lookup offset. + */ + /* Notice that max_null_row is max array index, we need count, so +1. */ + if (bitmap_init(&null_key, NULL, max_null_row + 1, FALSE)) + return TRUE; + + cur_key_idx= HA_POS_ERROR; + + return FALSE; +} + + +/* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + + @retval -1 + @retval 0 + @retval +1 +*/ + +int +Ordered_key::cmp_keys_by_row_data(ha_rows a, ha_rows b) +{ + uchar *rowid_a, *rowid_b; + int error, cmp_res; + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + + if (a == b) + return 0; + /* Get the corresponding rowids. */ + rowid_a= row_num_to_rowid + a * rowid_length; + rowid_b= row_num_to_rowid + b * rowid_length; + /* Fetch the rows for comparison. */ + error= tbl->file->ha_rnd_pos(tbl->record[0], rowid_a); + DBUG_ASSERT(!error); + error= tbl->file->ha_rnd_pos(tbl->record[1], rowid_b); + DBUG_ASSERT(!error); + /* + Compare the two rows by the corresponding values of the indexed + columns. + */ + for (uint i= 0; i < key_column_count; i++) + { + Field *cur_field= key_columns[i]->field; + if ((cmp_res= cur_field->cmp_offset(tbl->s->rec_buff_length))) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +int +Ordered_key::cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b) +{ + /* The result of comparing the two keys according to their row data. */ + int cmp_row_res= key->cmp_keys_by_row_data(*a, *b); + if (cmp_row_res) + return cmp_row_res; + return (*a < *b) ? -1 : (*a > *b) ? 1 : 0; +} + + +void Ordered_key::sort_keys() +{ + my_qsort2(key_buff, key_buff_elements, sizeof(rownum_t), + (qsort2_cmp) &cmp_keys_by_row_data_and_rownum, (void*) this); + /* Invalidate the current row position. */ + cur_key_idx= HA_POS_ERROR; +} + + +/* + The fraction of rows that do not contain NULL in the columns indexed by + this key. + + @retval 1 if there are no NULLs + @retval 0 if only NULLs +*/ + +double Ordered_key::null_selectivity() +{ + /* We should not be processing empty tables. */ + DBUG_ASSERT(tbl->file->stats.records); + return (1 - (double) null_count / (double) tbl->file->stats.records); +} + + +/* + Compare the value(s) of the current key in 'search_key' with the + data of the current table record. + + @notes The comparison result follows from the way compare_pred + is created in Ordered_key::init. Currently compare_pred compares + a field in of the current row with the corresponding Item that + contains the search key. + + @param row_num Number of the row (not index in the key_buff array) + + @retval -1 if (current row < search_key) + @retval 0 if (current row == search_key) + @retval +1 if (current row > search_key) +*/ + +int Ordered_key::cmp_key_with_search_key(rownum_t row_num) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + uchar *cur_rowid= row_num_to_rowid + row_num * rowid_length; + int error, cmp_res; + + error= tbl->file->ha_rnd_pos(tbl->record[0], cur_rowid); + DBUG_ASSERT(!error); + + for (uint i= 0; i < key_column_count; i++) + { + cmp_res= compare_pred[i]->get_comparator()->compare(); + /* Unlike Arg_comparator::compare_row() here there should be no NULLs. */ + DBUG_ASSERT(!compare_pred[i]->null_value); + if (cmp_res) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +/* + Find a key in a sorted array of keys via binary search. + + see create_subq_in_equalities() +*/ + +bool Ordered_key::lookup() +{ + DBUG_ASSERT(key_buff_elements); + + ha_rows lo= 0; + ha_rows hi= key_buff_elements - 1; + ha_rows mid; + int cmp_res; + + while (lo <= hi) + { + mid= lo + (hi - lo) / 2; + cmp_res= cmp_key_with_search_key(key_buff[mid]); + /* + In order to find the minimum match, check if the pevious element is + equal or smaller than the found one. If equal, we need to search further + to the left. + */ + if (!cmp_res && mid > 0) + cmp_res= !cmp_key_with_search_key(key_buff[mid - 1]) ? 1 : 0; + + if (cmp_res == -1) + { + /* row[mid] < search_key */ + lo= mid + 1; + } + else if (cmp_res == 1) + { + /* row[mid] > search_key */ + if (!mid) + goto not_found; + hi= mid - 1; + } + else + { + /* row[mid] == search_key */ + cur_key_idx= mid; + return TRUE; + } + } +not_found: + cur_key_idx= HA_POS_ERROR; + return FALSE; +} + + +/* + Move the current index pointer to the next key with the same column + values as the current key. Since the index is sorted, all such keys + are contiguous. +*/ + +bool Ordered_key::next_same() +{ + DBUG_ASSERT(key_buff_elements); + + if (cur_key_idx < key_buff_elements - 1) + { + /* + TIMOUR: + The below is quite inefficient, since as a result we will fetch every + row (except the last one) twice. There must be a more efficient way, + e.g. swapping record[0] and record[1], and reading only the new record. + */ + if (!cmp_keys_by_row_data(key_buff[cur_key_idx], key_buff[cur_key_idx + 1])) + { + ++cur_key_idx; + return TRUE; + } + } + return FALSE; +} + + +void Ordered_key::print(String *str) +{ + uint i; + str->append("{idx="); + str->qs_append(keyid); + str->append(", ("); + for (i= 0; i < key_column_count - 1; i++) + { + str->append(key_columns[i]->field->field_name); + str->append(", "); + } + str->append(key_columns[i]->field->field_name); + str->append("), "); + + str->append("null_bitmap: (bits="); + str->qs_append(null_key.n_bits); + str->append(", nulls= "); + str->qs_append((double)null_count); + str->append(", min_null= "); + str->qs_append((double)min_null_row); + str->append(", max_null= "); + str->qs_append((double)max_null_row); + str->append("), "); + + str->append('}'); +} + + +subselect_partial_match_engine::subselect_partial_match_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_engine(item_arg, result_arg), + tmp_table(tmp_table_arg), lookup_engine(engine_arg), + equi_join_conds(equi_join_conds_arg), + covering_null_row_width(covering_null_row_width_arg) +{} + + +int subselect_partial_match_engine::exec() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + int res; + + /* Try to find a matching row by index lookup. */ + res= lookup_engine->copy_ref_key_simple(); + if (res == -1) + { + /* The result is FALSE based on the outer reference. */ + item_in->value= 0; + item_in->null_value= 0; + return 0; + } + else if (res == 0) + { + /* Search for a complete match. */ + if ((res= lookup_engine->index_lookup())) + { + /* An error occured during lookup(). */ + item_in->value= 0; + item_in->null_value= 0; + return res; + } + else if (item_in->value) + { + /* + A complete match was found, the result of IN is TRUE. + Notice: (this->item == lookup_engine->item) + */ + return 0; + } + } + + if (covering_null_row_width == tmp_table->s->fields) + { + /* + If there is a NULL-only row that coveres all columns the result of IN + is UNKNOWN. + */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + return 0; + } + + /* + There is no complete match. Look for a partial match (UNKNOWN result), or + no match (FALSE). + */ + if (tmp_table->file->inited) + tmp_table->file->ha_index_end(); + + if (partial_match()) + { + /* The result of IN is UNKNOWN. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + } + else + { + /* The result of IN is FALSE. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 0; + item_in->null_value= 0; + } + + return 0; +} + + +void subselect_partial_match_engine::print(String *str, + enum_query_type query_type) +{ + /* + Should never be called as the actual engine cannot be known at query + optimization time. + */ + DBUG_ASSERT(FALSE); +} + + +/* + @param non_null_key_parts + @param partial_match_key_parts A union of all single-column NULL key parts. + @param count_partial_match_columns Number of NULL keyparts (set bits above). + + @retval FALSE the engine was initialized successfully + @retval TRUE there was some (memory allocation) error during initialization, + such errors should be interpreted as revert to other strategy +*/ + +bool +subselect_rowid_merge_engine::init(MY_BITMAP *non_null_key_parts, + MY_BITMAP *partial_match_key_parts) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tmp_table->file->ref_length; + ha_rows row_count= tmp_table->file->stats.records; + rownum_t cur_rownum= 0; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + uint cur_keyid= 0; + Item_in_subselect *item_in= (Item_in_subselect*) item; + int error; + + if (keys_count == 0) + { + /* There is nothing to initialize, we will only do regular lookups. */ + return FALSE; + } + + DBUG_ASSERT(!covering_null_row_width || (covering_null_row_width && + keys_count == 1 && + non_null_key_parts)); + /* + Allocate buffers to hold the merged keys and the mapping between rowids and + row numbers. + */ + if (!(merge_keys= (Ordered_key**) thd->alloc(keys_count * + sizeof(Ordered_key*))) || + !(row_num_to_rowid= (uchar*) my_malloc(row_count * rowid_length * + sizeof(uchar), MYF(MY_WME)))) + return TRUE; + + /* Create the only non-NULL key if there is any. */ + if (non_null_key_parts) + { + non_null_key= new Ordered_key(cur_keyid, tmp_table, item_in->left_expr, + 0, 0, 0, row_num_to_rowid); + if (non_null_key->init(non_null_key_parts)) + return TRUE; + merge_keys[cur_keyid]= non_null_key; + merge_keys[cur_keyid]->first(); + ++cur_keyid; + } + + /* + If there is a covering NULL row, the only key that is needed is the + only non-NULL key that is already created above. We create keys on + NULL-able columns only if there is no covering NULL row. + */ + if (!covering_null_row_width) + { + if (bitmap_init_memroot(&matching_keys, keys_count, thd->mem_root) || + bitmap_init_memroot(&matching_outer_cols, keys_count, thd->mem_root) || + bitmap_init_memroot(&null_only_columns, keys_count, thd->mem_root)) + return TRUE; + + /* + Create one single-column NULL-key for each column in + partial_match_key_parts. + */ + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i)) + continue; + + if (result_sink->get_null_count_of_col(i) == row_count) + bitmap_set_bit(&null_only_columns, cur_keyid); + else + { + merge_keys[cur_keyid]= new Ordered_key( + cur_keyid, tmp_table, + item_in->left_expr->element_index(i), + result_sink->get_null_count_of_col(i), + result_sink->get_min_null_of_col(i), + result_sink->get_max_null_of_col(i), + row_num_to_rowid); + if (merge_keys[cur_keyid]->init(i)) + return TRUE; + merge_keys[cur_keyid]->first(); + } + ++cur_keyid; + } + } + + /* Populate the indexes with data from the temporary table. */ + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + tmp_table->null_row= 0; + while (TRUE) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error == HA_ERR_RECORD_DELETED) + { + /* We get this for duplicate records that should not be in tmp_table. */ + continue; + } + /* + This is a temp table that we fully own, there should be no other + cause to stop the iteration than EOF. + */ + DBUG_ASSERT(!error || error == HA_ERR_END_OF_FILE); + if (error == HA_ERR_END_OF_FILE) + { + DBUG_ASSERT(cur_rownum == tmp_table->file->stats.records); + break; + } + + /* + Save the position of this record in the row_num -> rowid mapping. + */ + tmp_table->file->position(tmp_table->record[0]); + memcpy(row_num_to_rowid + cur_rownum * rowid_length, + tmp_table->file->ref, rowid_length); + + /* Add the current row number to the corresponding keys. */ + if (non_null_key) + { + /* By definition there are no NULLs in the non-NULL key. */ + non_null_key->add_key(cur_rownum); + } + + for (uint i= (non_null_key ? 1 : 0); i < keys_count; i++) + { + /* + Check if the first and only indexed column contains NULL in the curent + row, and add the row number to the corresponding key. + */ + if (tmp_table->field[merge_keys[i]->get_field_idx(0)]->is_null()) + merge_keys[i]->set_null(cur_rownum); + else + merge_keys[i]->add_key(cur_rownum); + } + ++cur_rownum; + } + + tmp_table->file->ha_rnd_end(); + + /* Sort all the keys by their NULL selectivity. */ + my_qsort(merge_keys, keys_count, sizeof(Ordered_key*), + (qsort_cmp) cmp_keys_by_null_selectivity); + + /* Sort the keys in each of the indexes. */ + for (uint i= 0; i < keys_count; i++) + merge_keys[i]->sort_keys(); + + if (init_queue(&pq, keys_count, 0, FALSE, + subselect_rowid_merge_engine::cmp_keys_by_cur_rownum, NULL)) + return TRUE; + + return FALSE; +} + + +subselect_rowid_merge_engine::~subselect_rowid_merge_engine() +{ + /* None of the resources below is allocated if there are no ordered keys. */ + if (keys_count) + { + my_free((char*) row_num_to_rowid, MYF(0)); + for (uint i= 0; i < keys_count; i++) + delete merge_keys[i]; + delete_queue(&pq); + if (tmp_table->file->inited == handler::RND) + tmp_table->file->ha_rnd_end(); + } +} + + +void subselect_rowid_merge_engine::cleanup() +{ +} + + +/* + Quick sort comparison function to compare keys in order of decreasing bitmap + selectivity, so that the most selective keys come first. + + @param k1 first key to compare + @param k2 second key to compare + + @retval 1 if k1 is less selective than k2 + @retval 0 if k1 is equally selective as k2 + @retval -1 if k1 is more selective than k2 +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_null_selectivity(Ordered_key **k1, + Ordered_key **k2) +{ + double k1_sel= (*k1)->null_selectivity(); + double k2_sel= (*k2)->null_selectivity(); + if (k1_sel < k2_sel) + return 1; + if (k1_sel > k2_sel) + return -1; + return 0; +} + + +/* +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_cur_rownum(void *arg, + uchar *k1, uchar *k2) +{ + rownum_t r1= ((Ordered_key*) k1)->current(); + rownum_t r2= ((Ordered_key*) k2)->current(); + + return (r1 < r2) ? -1 : (r1 > r2) ? 1 : 0; +} + + +/* + Check if certain table row contains a NULL in all columns for which there is + no match in the corresponding value index. + + @retval TRUE if a NULL row exists + @retval FALSE otherwise +*/ + +bool subselect_rowid_merge_engine::test_null_row(rownum_t row_num) +{ + Ordered_key *cur_key; + uint cur_id; + for (uint i = 0; i < keys_count; i++) + { + cur_key= merge_keys[i]; + cur_id= cur_key->get_keyid(); + if (bitmap_is_set(&matching_keys, cur_id)) + { + /* + The key 'i' (with id 'cur_keyid') already matches a value in row 'row_num', + thus we skip it as it can't possibly match a NULL. + */ + continue; + } + if (!cur_key->is_null(row_num)) + return FALSE; + } + return TRUE; +} + + +/* + @retval TRUE there is a partial match (UNKNOWN) + @retval FALSE there is no match at all (FALSE) +*/ + +bool subselect_rowid_merge_engine::partial_match() +{ + Ordered_key *min_key; /* Key that contains the current minimum position. */ + rownum_t min_row_num; /* Current row number of min_key. */ + Ordered_key *cur_key; + rownum_t cur_row_num; + uint count_nulls_in_search_key= 0; + bool res= FALSE; + + /* If there is a non-NULL key, it must be the first key in the keys array. */ + DBUG_ASSERT(!non_null_key || (non_null_key && merge_keys[0] == non_null_key)); + + /* All data accesses during execution are via handler::ha_rnd_pos() */ + tmp_table->file->ha_rnd_init(0); + + /* Check if there is a match for the columns of the only non-NULL key. */ + if (non_null_key && !non_null_key->lookup()) + { + res= FALSE; + goto end; + } + + /* + If there is a NULL (sub)row that covers all NULL-able columns, + then there is a guranteed partial match, and we don't need to search + for the matching row. + */ + if (covering_null_row_width) + { + res= TRUE; + goto end; + } + + if (non_null_key) + queue_insert(&pq, (uchar *) non_null_key); + /* + Do not add the non_null_key, since it was already processed above. + */ + bitmap_clear_all(&matching_outer_cols); + for (uint i= test(non_null_key); i < keys_count; i++) + { + DBUG_ASSERT(merge_keys[i]->get_column_count() == 1); + if (merge_keys[i]->get_search_key(0)->is_null()) + { + ++count_nulls_in_search_key; + bitmap_set_bit(&matching_outer_cols, merge_keys[i]->get_keyid()); + } + else if (merge_keys[i]->lookup()) + queue_insert(&pq, (uchar *) merge_keys[i]); + } + + /* + If the outer reference consists of only NULLs, or if it has NULLs in all + nullable columns, the result is UNKNOWN. + */ + if (count_nulls_in_search_key == + ((Item_in_subselect *) item)->left_expr->cols() - + (non_null_key ? non_null_key->get_column_count() : 0)) + { + res= TRUE; + goto end; + } + + /* + If there is no NULL (sub)row that covers all NULL columns, and there is no + single match for any of the NULL columns, the result is FALSE. + */ + if (pq.elements - test(non_null_key) == 0) + { + res= FALSE; + goto end; + } + + DBUG_ASSERT(pq.elements); + + min_key= (Ordered_key*) queue_remove(&pq, 0); + min_row_num= min_key->current(); + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + if (min_key->next_same()) + queue_insert(&pq, (uchar *) min_key); + + if (pq.elements == 0) + { + /* + Check the only matching row of the only key min_key for NULL matches + in the other columns. + */ + res= test_null_row(min_row_num); + goto end; + } + + while (TRUE) + { + cur_key= (Ordered_key*) queue_remove(&pq, 0); + cur_row_num= cur_key->current(); + + if (cur_row_num == min_row_num) + bitmap_set_bit(&matching_keys, cur_key->get_keyid()); + else + { + /* Follows from the correct use of priority queue. */ + DBUG_ASSERT(cur_row_num > min_row_num); + if (test_null_row(min_row_num)) + { + res= TRUE; + goto end; + } + else + { + min_key= cur_key; + min_row_num= cur_row_num; + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + } + } + + if (cur_key->next_same()) + queue_insert(&pq, (uchar *) cur_key); + + if (pq.elements == 0) + { + /* Check the last row of the last column in PQ for NULL matches. */ + res= test_null_row(min_row_num); + goto end; + } + } + + /* We should never get here - all branches must be handled explicitly above. */ + DBUG_ASSERT(FALSE); + +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +subselect_table_scan_engine::subselect_table_scan_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg) +{} + + +/* + TIMOUR: + This method is based on subselect_uniquesubquery_engine::scan_table(). + Consider refactoring somehow, 80% of the code is the same. + + for each row_i in tmp_table + { + count_matches= 0; + for each row element row_i[j] + { + if (outer_ref[j] is NULL || row_i[j] is NULL || outer_ref[j] == row_i[j]) + ++count_matches; + } + if (count_matches == outer_ref.elements) + return TRUE + } + return FALSE +*/ + +bool subselect_table_scan_engine::partial_match() +{ + List_iterator_fast<Item> equality_it(*equi_join_conds); + Item *cur_eq; + uint count_matches; + int error; + bool res; + + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + /* + TIMOUR: + scan_table() also calls "table->null_row= 0;", why, do we need it? + */ + for (;;) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(tmp_table, error); + break; + } + } + + equality_it.rewind(); + count_matches= 0; + while ((cur_eq= equality_it++)) + { + DBUG_ASSERT(cur_eq->type() == Item::FUNC_ITEM && + ((Item_func*)cur_eq)->functype() == Item_func::EQ_FUNC); + if (!cur_eq->val_int() && !cur_eq->null_value) + break; + ++count_matches; + } + if (count_matches == tmp_table->s->fields) + { + res= TRUE; /* Found a matching row. */ + goto end; + } + } + + res= FALSE; +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +void subselect_table_scan_engine::cleanup() +{ +} === modified file 'sql/item_subselect.h' --- a/sql/item_subselect.h 2010-02-11 23:59:58 +0000 +++ b/sql/item_subselect.h 2010-03-09 10:14:06 +0000 @@ -297,7 +297,7 @@ public: Representation of IN subquery predicates of the form "left_expr IN (SELECT ...)". - @detail + @details This class has: - A "subquery execution engine" (as a subclass of Item_subselect) that allows it to evaluate subqueries. (and this class participates in execution by @@ -319,6 +319,12 @@ protected: */ List<Cached_item> *left_expr_cache; bool first_execution; + /* + Set to TRUE if at query execution time we determine that this item's + value is a constant during this execution. We need this member because + it is not possible to substitute 'this' with a constant item. + */ + bool is_constant; /* expr & optimizer used in subselect rewriting to store Item for @@ -387,8 +393,8 @@ public: Item_in_subselect(Item * left_expr, st_select_lex *select_lex); Item_in_subselect() :Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), abort_on_null(0), pushed_cond_guards(NULL), - exec_method(NOT_TRANSFORMED), upper_item(0) + is_constant(FALSE), optimizer(0), abort_on_null(0), + pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), upper_item(0) {} void cleanup(); subs_type substype() { return IN_SUBS; } @@ -421,6 +427,8 @@ public: void update_used_tables(); bool setup_engine(); bool init_left_expr_cache(); + /* Inform 'this' that it was computed, and contains a valid result. */ + void set_first_execution() { if (first_execution) first_execution= FALSE; } bool is_expensive_processor(uchar *arg); friend class Item_ref_null_helper; @@ -428,6 +436,7 @@ public: friend class Item_in_optimizer; friend class subselect_indexsubquery_engine; friend class subselect_hash_sj_engine; + friend class subselect_partial_match_engine; }; @@ -462,7 +471,8 @@ public: enum enum_engine_type {ABSTRACT_ENGINE, SINGLE_SELECT_ENGINE, UNION_ENGINE, UNIQUESUBQUERY_ENGINE, - INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE}; + INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE, + ROWID_MERGE_ENGINE, TABLE_SCAN_ENGINE}; subselect_engine(Item_subselect *si, select_result_interceptor *res) :thd(0) @@ -635,8 +645,10 @@ public: virtual void print (String *str, enum_query_type query_type); bool change_result(Item_subselect *si, select_result_interceptor *result); bool no_tables(); + int index_lookup(); /* TIMOUR: this method needs refactoring. */ int scan_table(); bool copy_ref_key(); + int copy_ref_key_simple(); /* TIMOUR: this method needs refactoring. */ bool no_rows() { return empty_result_set; } virtual enum_engine_type engine_type() { return UNIQUESUBQUERY_ENGINE; } }; @@ -705,50 +717,439 @@ inline bool Item_subselect::is_uncacheab /** - Compute an IN predicate via a hash semi-join. The subquery is materialized - during the first evaluation of the IN predicate. The IN predicate is executed - via the functionality inherited from subselect_uniquesubquery_engine. + Compute an IN predicate via a hash semi-join. This class is responsible for + the materialization of the subquery, and the selection of the correct and + optimal execution method (e.g. direct index lookup, or partial matching) for + the IN predicate. */ -class subselect_hash_sj_engine: public subselect_uniquesubquery_engine +class subselect_hash_sj_engine : public subselect_engine { protected: + /* The table into which the subquery is materialized. */ + TABLE *tmp_table; /* TRUE if the subquery was materialized into a temp table. */ bool is_materialized; /* The old engine already chosen at parse time and stored in permanent memory. Through this member we can re-create and re-prepare materialize_join for - each execution of a prepared statement. We akso resuse the functionality + each execution of a prepared statement. We also reuse the functionality of subselect_single_select_engine::[prepare | cols]. */ subselect_single_select_engine *materialize_engine; + /* The engine used to compute the IN predicate. */ + subselect_engine *lookup_engine; /* QEP to execute the subquery and materialize its result into a temporary table. Created during the first call to exec(). */ JOIN *materialize_join; - /* Temp table context of the outer select's JOIN. */ - TMP_TABLE_PARAM *tmp_param; + + /* Keyparts of the only non-NULL composite index in a rowid merge. */ + MY_BITMAP non_null_key_parts; + /* Keyparts of the single column indexes with NULL, one keypart per index. */ + MY_BITMAP partial_match_key_parts; + uint count_partial_match_columns; + uint count_null_only_columns; + /* + A conjunction of all the equality condtions between all pairs of expressions + that are arguments of an IN predicate. We need these to post-filter some + IN results because index lookups sometimes match values that are actually + not equal to the search key in SQL terms. + */ + Item_cond_and *semi_join_conds; + /* Possible execution strategies that can be used to compute hash semi-join.*/ + enum exec_strategy { + UNDEFINED, + COMPLETE_MATCH, /* Use regular index lookups. */ + PARTIAL_MATCH, /* Use some partial matching strategy. */ + PARTIAL_MATCH_MERGE, /* Use partial matching through index merging. */ + PARTIAL_MATCH_SCAN, /* Use partial matching through table scan. */ + IMPOSSIBLE /* Subquery materialization is not applicable. */ + }; + /* The chosen execution strategy. Computed after materialization. */ + exec_strategy strategy; +protected: + exec_strategy get_strategy_using_schema(); + exec_strategy get_strategy_using_data(); + size_t rowid_merge_buff_size(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + void choose_partial_match_strategy(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + bool make_semi_join_conds(); + subselect_uniquesubquery_engine* make_unique_engine(); public: subselect_hash_sj_engine(THD *thd, Item_subselect *in_predicate, - subselect_single_select_engine *old_engine) - :subselect_uniquesubquery_engine(thd, NULL, in_predicate, NULL), - is_materialized(FALSE), materialize_engine(old_engine), - materialize_join(NULL), tmp_param(NULL) - {} + subselect_single_select_engine *old_engine) + :subselect_engine(in_predicate, NULL), tmp_table(NULL), + is_materialized(FALSE), materialize_engine(old_engine), lookup_engine(NULL), + materialize_join(NULL), count_partial_match_columns(0), + count_null_only_columns(0), semi_join_conds(NULL), strategy(UNDEFINED) + { + set_thd(thd); + } ~subselect_hash_sj_engine(); bool init_permanent(List<Item> *tmp_columns); bool init_runtime(); void cleanup(); - int prepare() { return 0; } + int prepare() { return 0; } /* Override virtual function in base class. */ int exec(); - virtual void print (String *str, enum_query_type query_type); + virtual void print(String *str, enum_query_type query_type); uint cols() { return materialize_engine->cols(); } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + table_map upper_select_const_tables() { return 0; } + bool no_rows() { return !tmp_table->file->stats.records; } virtual enum_engine_type engine_type() { return HASH_SJ_ENGINE; } + /* + TODO: factor out all these methods in a base subselect_index_engine class + because all of them have dummy implementations and should never be called. + */ + void fix_length_and_dec(Item_cache** row);//=>base class + void exclude(); //=>base class + //=>base class + bool change_result(Item_subselect *si, select_result_interceptor *result); + bool no_tables();//=>base class +}; + + +/* + Distinguish the type od (0-based) row numbers from the type of the index into + an array of row numbers. +*/ +typedef ha_rows rownum_t; + + +/* + An Ordered_key is an in-memory table index that allows O(log(N)) time + lookups of a multi-part key. + + If the index is over a single column, then this column may contain NULLs, and + the NULLs are stored and tested separately for NULL in O(1) via is_null(). + Multi-part indexes assume that the indexed columns do not contain NULLs. + + TODO: + = Due to the unnatural assymetry between single and multi-part indexes, it + makes sense to somehow refactor or extend the class. + + = This class can be refactored into a base abstract interface, and two + subclasses: + - one to represent single-column indexes, and + - another to represent multi-column indexes. + Such separation would allow slightly more efficient implementation of + the single-column indexes. + = The current design requires such indexes to be fully recreated for each + PS (re)execution, however most of the comprising objects can be reused. +*/ + +class Ordered_key : public Sql_alloc +{ +protected: + /* + Index of the key in an array of keys. This index allows to + construct (sub)sets of keys represented by bitmaps. + */ + uint keyid; + /* The table being indexed. */ + TABLE *tbl; + /* The columns being indexed. */ + Item_field **key_columns; + /* Number of elements in 'key_columns' (number of key parts). */ + uint key_column_count; + /* + An expression, or sequence of expressions that forms the search key. + The search key is a sequence when it is Item_row. Each element of the + sequence is accessible via Item::element_index(int i). + */ + Item *search_key; + +/* Value index related members. */ + /* + The actual value index, consists of a sorted sequence of row numbers. + */ + rownum_t *key_buff; + /* Number of elements in key_buff. */ + ha_rows key_buff_elements; + /* Current element in 'key_buff'. */ + ha_rows cur_key_idx; + /* + Mapping from row numbers to row ids. The element row_num_to_rowid[i] + contains a buffer with the rowid for the row numbered 'i'. + The memory for this member is not maintanined by this class because + all Ordered_key indexes of the same table share the same mapping. + */ + uchar *row_num_to_rowid; + /* + A sequence of predicates to compare the search key with the corresponding + columns of a table row from the index. + */ + Item_func_lt **compare_pred; + +/* Null index related members. */ + MY_BITMAP null_key; + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + +protected: + bool alloc_keys_buffers(); + /* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + */ + int cmp_keys_by_row_data(rownum_t a, rownum_t b); + static int cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b); + + int cmp_key_with_search_key(rownum_t row_num); + +public: + Ordered_key(uint keyid_arg, TABLE *tbl_arg, + Item *search_key_arg, ha_rows null_count_arg, + ha_rows min_null_row_arg, ha_rows max_null_row_arg, + uchar *row_num_to_rowid_arg); + ~Ordered_key(); + void cleanup(); + /* Initialize a multi-column index. */ + bool init(MY_BITMAP *columns_to_index); + /* Initialize a single-column index. */ + bool init(int col_idx); + + uint get_column_count() { return key_column_count; } + uint get_keyid() { return keyid; } + uint get_field_idx(uint i) + { + DBUG_ASSERT(i < key_column_count); + return key_columns[i]->field->field_index; + } + /* + Get the search key element that corresponds to the i-th key part of this + index. + */ + Item *get_search_key(uint i) + { + return search_key->element_index(key_columns[i]->field->field_index); + } + void add_key(rownum_t row_num) + { + /* The caller must know how many elements to add. */ + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + key_buff[cur_key_idx]= row_num; + ++cur_key_idx; + } + + void sort_keys(); + double null_selectivity(); + + /* + Position the current element at the first row that matches the key. + The key itself is propagated by evaluating the current value(s) of + this->search_key. + */ + bool lookup(); + /* Move the current index cursor to the first key. */ + void first() + { + DBUG_ASSERT(key_buff_elements); + cur_key_idx= 0; + } + /* TODO */ + bool next_same(); + /* Move the current index cursor to the next key. */ + bool next() + { + DBUG_ASSERT(key_buff_elements); + if (cur_key_idx < key_buff_elements - 1) + { + ++cur_key_idx; + return TRUE; + } + return FALSE; + }; + /* Return the current index element. */ + rownum_t current() + { + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + return key_buff[cur_key_idx]; + } + + void set_null(rownum_t row_num) + { + bitmap_set_bit(&null_key, row_num); + } + bool is_null(rownum_t row_num) + { + /* + Indexes consisting of only NULLs do not have a bitmap buffer at all. + Their only initialized member is 'n_bits', which is equal to the number + of temp table rows. + */ + if (null_count == tbl->file->stats.records) + { + DBUG_ASSERT(tbl->file->stats.records == null_key.n_bits); + return TRUE; + } + if (row_num > max_null_row || row_num < min_null_row) + return FALSE; + return bitmap_is_set(&null_key, row_num); + } + void print(String *str); +}; + + +class subselect_partial_match_engine : public subselect_engine +{ +protected: + /* The temporary table that contains a materialized subquery. */ + TABLE *tmp_table; + /* + The engine used to check whether an IN predicate is TRUE or not. If not + TRUE, then subselect_rowid_merge_engine further distinguishes between + FALSE and UNKNOWN. + */ + subselect_uniquesubquery_engine *lookup_engine; + /* A list of equalities between each pair of IN operands. */ + List<Item> *equi_join_conds; + /* + If there is a row, such that all its NULL-able components are NULL, this + member is set to the number of covered columns. If there is no covering + row, then this is 0. + */ + uint covering_null_row_width; +protected: + virtual bool partial_match()= 0; +public: + subselect_partial_match_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + int prepare() { return 0; } + int exec(); + void fix_length_and_dec(Item_cache**) {} + uint cols() { /* TODO: what is the correct value? */ return 1; } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + void exclude() {} + table_map upper_select_const_tables() { return 0; } + bool change_result(Item_subselect*, select_result_interceptor*) + { DBUG_ASSERT(FALSE); return false; } + bool no_tables() { return false; } + bool no_rows() + { + /* + TODO: It is completely unclear what is the semantics of this + method. The current result is computed so that the call to no_rows() + from Item_in_optimizer::val_int() sets Item_in_optimizer::null_value + correctly. + */ + return !(((Item_in_subselect *) item)->null_value); + } + void print(String*, enum_query_type); + + friend void subselect_hash_sj_engine::cleanup(); +}; + + +class subselect_rowid_merge_engine: public subselect_partial_match_engine +{ +protected: + /* + Mapping from row numbers to row ids. The rowids are stored sequentially + in the array - rowid[i] is located in row_num_to_rowid + i * rowid_length. + */ + uchar *row_num_to_rowid; + /* + A subset of all the keys for which there is a match for the same row. + Used during execution. Computed for each outer reference + */ + MY_BITMAP matching_keys; + /* + The columns of the outer reference that are NULL. Computed for each + outer reference. + */ + MY_BITMAP matching_outer_cols; + /* + Columns that consist of only NULLs. Such columns match any value. + Computed once per query execution. + */ + MY_BITMAP null_only_columns; + /* + Indexes of row numbers, sorted by <column_value, row_number>. If an + index may contain NULLs, the NULLs are stored efficiently in a bitmap. + + The indexes are sorted by the selectivity of their NULL sub-indexes, the + one with the fewer NULLs is first. Thus, if there is any index on + non-NULL columns, it is contained in keys[0]. + */ + Ordered_key **merge_keys; + /* The number of elements in keys. */ + uint keys_count; + /* + An index on all non-NULL columns of 'tmp_table'. The index has the + logical form: <[v_i1 | ... | v_ik], rownum>. It allows to find the row + number where the columns c_i1,...,c1_k contain the values v_i1,...,v_ik. + If such an index exists, it is always the first element of 'keys'. + */ + Ordered_key *non_null_key; + /* + Priority queue of Ordered_key indexes, one per NULLable column. + This queue is used by the partial match algorithm in method exec(). + */ + QUEUE pq; +protected: + /* + Comparison function to compare keys in order of decreasing bitmap + selectivity. + */ + static int cmp_keys_by_null_selectivity(Ordered_key **k1, Ordered_key **k2); + /* + Comparison function used by the priority queue pq, the 'smaller' key + is the one with the smaller current row number. + */ + static int cmp_keys_by_cur_rownum(void *arg, uchar *k1, uchar *k2); + + bool test_null_row(rownum_t row_num); + bool partial_match(); +public: + subselect_rowid_merge_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, uint keys_count_arg, + uint covering_null_row_width_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg), + keys_count(keys_count_arg), non_null_key(NULL) + { + thd= lookup_engine->get_thd(); + } + ~subselect_rowid_merge_engine(); + bool init(MY_BITMAP *non_null_key_parts, MY_BITMAP *partial_match_key_parts); + void cleanup(); + virtual enum_engine_type engine_type() { return ROWID_MERGE_ENGINE; } }; + +class subselect_table_scan_engine: public subselect_partial_match_engine +{ +protected: + bool partial_match(); +public: + subselect_table_scan_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + void cleanup(); + virtual enum_engine_type engine_type() { return TABLE_SCAN_ENGINE; } +}; === modified file 'sql/mysql_priv.h' --- a/sql/mysql_priv.h 2010-01-17 14:55:08 +0000 +++ b/sql/mysql_priv.h 2010-03-09 10:14:06 +0000 @@ -552,12 +552,14 @@ protected: #define OPTIMIZER_SWITCH_LOOSE_SCAN 64 #define OPTIMIZER_SWITCH_MATERIALIZATION 128 #define OPTIMIZER_SWITCH_SEMIJOIN 256 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE 512 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN 1024 #ifdef DBUG_OFF -# define OPTIMIZER_SWITCH_LAST 512 +# define OPTIMIZER_SWITCH_LAST 2048 #else -# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 512 -# define OPTIMIZER_SWITCH_LAST 1024 +# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 2048 +# define OPTIMIZER_SWITCH_LAST 4096 #endif #ifdef DBUG_OFF @@ -570,8 +572,10 @@ protected: OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) -#else + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) +#else # define OPTIMIZER_SWITCH_DEFAULT (OPTIMIZER_SWITCH_INDEX_MERGE | \ OPTIMIZER_SWITCH_INDEX_MERGE_UNION | \ OPTIMIZER_SWITCH_INDEX_MERGE_SORT_UNION | \ @@ -581,7 +585,9 @@ protected: OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) #endif /* === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2010-01-17 14:55:08 +0000 +++ b/sql/mysqld.cc 2010-03-09 10:14:06 +0000 @@ -301,7 +301,9 @@ static const char *optimizer_switch_name "index_merge","index_merge_union","index_merge_sort_union", "index_merge_intersection", "index_condition_pushdown", - "firstmatch","loosescan","materialization", "semijoin", + "firstmatch","loosescan","materialization", "semijoin", + "partial_match_rowid_merge", + "partial_match_table_scan", #ifndef DBUG_OFF "table_elimination", #endif @@ -320,6 +322,8 @@ static const unsigned int optimizer_swit sizeof("loosescan") - 1, sizeof("materialization") - 1, sizeof("semijoin") - 1, + sizeof("partial_match_rowid_merge") - 1, + sizeof("partial_match_table_scan") - 1, #ifndef DBUG_OFF sizeof("table_elimination") - 1, #endif @@ -5794,7 +5798,8 @@ enum options_mysqld OPT_RECORD_RND_BUFFER, OPT_DIV_PRECINCREMENT, OPT_RELAY_LOG_SPACE_LIMIT, OPT_RELAY_LOG_PURGE, OPT_SLAVE_NET_TIMEOUT, OPT_SLAVE_COMPRESSED_PROTOCOL, OPT_SLOW_LAUNCH_TIME, - OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_DEBUGGING, OPT_DEBUG_FLUSH, + OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_ROWID_MERGE_BUFF_SIZE, + OPT_DEBUGGING, OPT_DEBUG_FLUSH, OPT_SORT_BUFFER, OPT_TABLE_OPEN_CACHE, OPT_TABLE_DEF_CACHE, OPT_THREAD_CONCURRENCY, OPT_THREAD_CACHE_SIZE, OPT_TMP_TABLE_SIZE, OPT_THREAD_STACK, @@ -7130,6 +7135,11 @@ The minimum value for this variable is 4 (uchar**) &max_system_variables.range_alloc_block_size, 0, GET_ULONG, REQUIRED_ARG, RANGE_ALLOC_BLOCK_SIZE, RANGE_ALLOC_BLOCK_SIZE, (longlong) ULONG_MAX, 0, 1024, 0}, + {"rowid_merge_buff_size", OPT_ROWID_MERGE_BUFF_SIZE, + "The size of the buffers used [NOT] IN evaluation via partial matching.", + (uchar**) &global_system_variables.rowid_merge_buff_size, + (uchar**) &max_system_variables.rowid_merge_buff_size, 0, GET_ULONG, + REQUIRED_ARG, 8*1024*1024L, 0, MAX_MEM_TABLE_SIZE/2, 0, 1, 0}, {"read_buffer_size", OPT_RECORD_BUFFER, "Each thread that does a sequential scan allocates a buffer of this size for each table it scans. If you do many sequential scans, you may want to increase this value.", (uchar**) &global_system_variables.read_buff_size, === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-15 06:32:54 +0000 +++ b/sql/opt_subselect.cc 2010-03-15 19:52:58 +0000 @@ -187,10 +187,10 @@ int check_and_do_in_subquery_rewrites(JO does not call setup_subquery_materialization(). We could make SELECT ... FROM DUAL call that function but that doesn't seem to be the case that is worth handling. - 4. Subquery predicate is a top-level predicate - (this implies it is not negated) - TODO: this is a limitation that should be lifted once we - implement correct NULL semantics (WL#3830) + 4. Either the subquery predicate is a top-level predicate, or at + least one partial match strategy is enabled. If no partial match + strategy is enabled, then materialization cannot be used for + non-top-level queries because it cannot handle NULLs correctly. 5. Subquery is non-correlated TODO: This is an overly restrictive condition. It can be extended to: @@ -204,8 +204,8 @@ int check_and_do_in_subquery_rewrites(JO (*) The subquery must be part of a SELECT statement. The current condition also excludes multi-table update statements. - We have to determine whether we will perform subquery materialization - before calling the IN=>EXISTS transformation, so that we know whether to + Determine whether we will perform subquery materialization before + calling the IN=>EXISTS transformation, so that we know whether to perform the whole transformation or only that part of it which wraps Item_in_subselect in an Item_in_optimizer. */ @@ -215,12 +215,14 @@ int check_and_do_in_subquery_rewrites(JO select_lex->master_unit()->first_select()->leaf_tables && // 3 thd->lex->sql_command == SQLCOM_SELECT && // * select_lex->outer_select()->leaf_tables && // 3A - subquery_types_allow_materialization(in_subs)) + subquery_types_allow_materialization(in_subs) && + // psergey-todo: duplicated_subselect_card_check: where it's done? + (in_subs->is_top_level_item() || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) &&//4 + !in_subs->is_correlated && // 5 + in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 { - // psergey-todo: duplicated_subselect_card_check: where it's done? - if (in_subs->is_top_level_item() && // 4 - !in_subs->is_correlated && // 5 - in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 in_subs->exec_method= Item_in_subselect::MATERIALIZATION; } === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-12-22 12:49:15 +0000 +++ b/sql/set_var.cc 2010-03-09 10:14:06 +0000 @@ -540,6 +540,9 @@ static sys_var_long_ptr sys_query_cache_ static sys_var_thd_ulong sys_range_alloc_block_size(&vars, "range_alloc_block_size", &SV::range_alloc_block_size); +static sys_var_thd_ulong sys_rowid_merge_buff_size(&vars, "rowid_merge_buff_size", + &SV::rowid_merge_buff_size); + static sys_var_thd_ulong sys_query_alloc_block_size(&vars, "query_alloc_block_size", &SV::query_alloc_block_size, 0, fix_thd_mem_root); === modified file 'sql/sql_class.cc' --- a/sql/sql_class.cc 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.cc 2010-02-19 21:55:57 +0000 @@ -42,6 +42,7 @@ #include "sp_rcontext.h" #include "sp_cache.h" +#include "sql_select.h" /* declares create_tmp_table() */ /* The following is used to initialise Table_ident with a internal @@ -2877,6 +2878,71 @@ bool select_dumpvar::send_eof() return 0; } + +bool +select_materialize_with_stats:: +create_result_table(THD *thd_arg, List<Item> *column_types, + bool is_union_distinct, ulonglong options, + const char *table_alias, bool bit_fields_as_long) +{ + DBUG_ASSERT(table == 0); + tmp_table_param.field_count= column_types->elements; + tmp_table_param.bit_fields_as_long= bit_fields_as_long; + + if (! (table= create_tmp_table(thd_arg, &tmp_table_param, *column_types, + (ORDER*) 0, is_union_distinct, 1, + options, HA_POS_ERROR, (char*) table_alias))) + return TRUE; + + col_stat= (Column_statistics*) table->in_use->alloc(table->s->fields * + sizeof(Column_statistics)); + if (!stat) + return TRUE; + + cleanup(); + + table->file->extra(HA_EXTRA_WRITE_CACHE); + table->file->extra(HA_EXTRA_IGNORE_DUP_KEY); + return FALSE; +} + + +/** + Override select_union::send_data to analyze each row for NULLs and to + update null_statistics before sending data to the client. + + @return TRUE if fatal error when sending data to the client + @return FALSE on success +*/ + +bool select_materialize_with_stats::send_data(List<Item> &items) +{ + List_iterator_fast<Item> item_it(items); + Item *cur_item; + Column_statistics *cur_col_stat= col_stat; + uint nulls_in_row= 0; + + ++count_rows; + + while ((cur_item= item_it++)) + { + if (cur_item->is_null()) + { + ++cur_col_stat->null_count; + cur_col_stat->max_null_row= count_rows; + if (!cur_col_stat->min_null_row) + cur_col_stat->min_null_row= count_rows; + ++nulls_in_row; + } + ++cur_col_stat; + } + if (nulls_in_row > max_nulls_in_row) + max_nulls_in_row= nulls_in_row; + + return select_union::send_data(items); +} + + /**************************************************************************** TMP_TABLE_PARAM ****************************************************************************/ === modified file 'sql/sql_class.h' --- a/sql/sql_class.h 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.h 2010-03-09 10:14:06 +0000 @@ -343,6 +343,8 @@ struct system_variables ulong mrr_buff_size; ulong div_precincrement; ulong sortbuff_size; + /* Total size of all buffers used by the subselect_rowid_merge_engine. */ + ulong rowid_merge_buff_size; ulong thread_handling; ulong tx_isolation; ulong completion_type; @@ -2740,19 +2742,20 @@ public: class select_union :public select_result_interceptor { +protected: TMP_TABLE_PARAM tmp_table_param; public: TABLE *table; - select_union() :table(0) {} + select_union() :table(0) { tmp_table_param.init(); } int prepare(List<Item> &list, SELECT_LEX_UNIT *u); bool send_data(List<Item> &items); bool send_eof(); bool flush(); - bool create_result_table(THD *thd, List<Item> *column_types, - bool is_distinct, ulonglong options, - const char *alias, bool bit_fields_as_long); + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); }; /* Base subselect interface class */ @@ -2776,6 +2779,74 @@ public: bool send_data(List<Item> &items); }; + +/* + This class specializes select_union to collect statistics about the + data stored in the temp table. Currently the class collects statistcs + about NULLs. +*/ + +class select_materialize_with_stats : public select_union +{ +protected: + class Column_statistics + { + public: + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + }; + + /* Array of statistics data per column. */ + Column_statistics* col_stat; + + /* + The number of columns in the biggest sub-row that consists of only + NULL values. + */ + ha_rows max_nulls_in_row; + /* + Count of rows writtent to the temp table. This is redundant as it is + already stored in handler::stats.records, however that one is relatively + expensive to compute (given we need that for evry row). + */ + ha_rows count_rows; + +public: + select_materialize_with_stats() {} + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); + bool init_result_table(ulonglong select_options); + bool send_data(List<Item> &items); + void cleanup() + { + memset(col_stat, 0, table->s->fields * sizeof(Column_statistics)); + max_nulls_in_row= 0; + count_rows= 0; + } + ha_rows get_null_count_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].null_count; + } + ha_rows get_max_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].max_null_row; + } + ha_rows get_min_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].min_null_row; + } + ha_rows get_max_nulls_in_row() { return max_nulls_in_row; } +}; + + /* used in independent ALL/ANY optimisation */ class select_max_min_finder_subselect :public select_subselect { === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-14 18:25:43 +0000 +++ b/sql/sql_select.cc 2010-03-15 19:52:58 +0000 @@ -874,6 +874,9 @@ JOIN::optimize() { DBUG_PRINT("info",("No tables")); error= 0; + /* Create all structures needed for materialized subquery execution. */ + if (setup_subquery_materialization()) + DBUG_RETURN(1); DBUG_RETURN(0); } error= -1; // Error is sent to client @@ -11258,7 +11261,7 @@ create_tmp_table(THD *thd,TMP_TABLE_PARA param->group_buff=group_buff; share->keys=1; share->uniques= test(using_unique_constraint); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME; keyinfo->usable_key_parts=keyinfo->key_parts= param->group_parts; @@ -11344,7 +11347,7 @@ create_tmp_table(THD *thd,TMP_TABLE_PARA keyinfo->key_parts * sizeof(KEY_PART_INFO)))) goto err; bzero((void*) key_part_info, keyinfo->key_parts * sizeof(KEY_PART_INFO)); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME | HA_NULL_ARE_EQUAL; keyinfo->key_length= 0; // Will compute the sum of the parts below.

1 0

[Maria-developers] Rev 2779: Merge in MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs in file:///home/psergey/dev/maria-5.3-subqueries-r7/
by Sergey Petrunya 15 Mar '10

15 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7/ ------------------------------------------------------------ revno: 2779 [merge] revision-id: psergey(a)askmonty.org-20100315150935-4xm838tskbh9k3ci parent: psergey(a)askmonty.org-20100315063535-jsp4jgya6lfqt8e6 parent: timour(a)sun.com-20100315143456-82d9rq3lbdscbr2n committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7 timestamp: Mon 2010-03-15 18:09:35 +0300 message: Merge in MWL#68: Subquery optimization: Efficient NOT IN execution with NULLs modified: mysql-test/include/mix1.inc sp1f-innodb_mysql.test-20060426055153-mgtahdmgajg7vffqbq4xrmkzbhvanlaz mysql-test/r/index_merge_myisam.result sp1f-index_merge_myisam.r-20060816114353-wd2664hjxwyjdvm4snup647av5fmxfln mysql-test/r/innodb_mysql.result sp1f-innodb_mysql.result-20060426055153-bychbbfnqtvmvrwccwhn24i6yi46uqjv mysql-test/r/myisam_mrr.result myisam_mrr.result-20091215071345-6wadxunod6vi8m48-1 mysql-test/r/ps.result sp1f-ps.result-20040405154119-efxzt5onloys45nfjak4gt44kr4awkdi mysql-test/r/subselect.result sp1f-subselect.result-20020512204640-zgegcsgavnfd7t7eyrf7ibuqomsw7uzo mysql-test/r/subselect3.result sp1f-subselect3.result-20061031174245-v7hvtc7uwevifiq4lziwv5gdcxpeak7t mysql-test/r/subselect3_jcl6.result subselect3_jcl6.resu-20100117143923-cf6j4mu5zzng00u7-1 mysql-test/r/subselect_no_mat.result subselect_no_mat.res-20100117143924-hut18sl9k2c7qdj8-1 mysql-test/r/subselect_no_opts.result subselect_no_opts.re-20100117143925-pabg7o8iyokjlu93-1 mysql-test/r/subselect_no_semijoin.result subselect_no_semijoi-20100117143925-9yfygtcm7fwsuq2p-1 mysql-test/r/subselect_sj.result subselect_sj.result-20100117143926-nrop4ku355g3kv8b-1 mysql-test/r/subselect_sj_jcl6.result subselect_sj_jcl6.re-20100117143928-7vzk51yaf29cdavp-1 mysql-test/t/ps.test sp1f-ps.test-20040405154119-4zqf6po44yypvz5foa2osprg5kb5ok63 mysql-test/t/subselect.test sp1f-subselect.test-20020512204640-lyqrayx6uwsn7zih6y7kerkenuitzbvr mysql-test/t/subselect3.test sp1f-subselect3.test-20061031174245-pcxt5ljylerxhx2jkfhrbqfv5vqcazlz sql/item_cmpfunc.h sp1f-item_cmpfunc.h-19700101030959-pcvbjplo4e4ng7ibynfhcd6pjyem57gr sql/item_subselect.cc sp1f-item_subselect.cc-20020512204640-qep43aqhsfrwkqmrobni6czc3fqj36oo sql/item_subselect.h sp1f-item_subselect.h-20020512204640-qdg77wil56cxyhtc2bjjdrppxq3wqgh3 sql/mysql_priv.h sp1f-mysql_priv.h-19700101030959-4fl65tqpop5zfgxaxkqotu2fa2ree5ci sql/mysqld.cc sp1f-mysqld.cc-19700101030959-zpswdvekpvixxzxf7gdtofzel7nywtfj sql/opt_subselect.cc opt_subselect.cc-20100215190428-nekkl8wisp0k6nlk-1 sql/set_var.cc sp1f-set_var.cc-20020723153119-nwbpg2pwpz55pfw7yfzaxt7hsszzy7y3 sql/sql_class.cc sp1f-sql_class.cc-19700101030959-rpotnweaff2pikkozh3butrf7mv3oero sql/sql_class.h sp1f-sql_class.h-19700101030959-jnqnbrjyqsvgncsibnumsmg3lyi7pa5s sql/sql_select.cc sp1f-sql_select.cc-19700101030959-egb7whpkh76zzvikycs5nsnuviu4fdlb === modified file 'mysql-test/include/mix1.inc' --- a/mysql-test/include/mix1.inc 2009-09-15 06:08:54 +0000 +++ b/mysql-test/include/mix1.inc 2010-03-11 21:43:31 +0000 @@ -1177,8 +1177,11 @@ create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; --echo End of 5.0 tests === modified file 'mysql-test/r/index_merge_myisam.result' --- a/mysql-test/r/index_merge_myisam.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/index_merge_myisam.result 2010-03-11 21:43:31 +0000 @@ -1419,19 +1419,19 @@ # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge=off,index_merge_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge_union=on'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,index_merge_sort_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=4; ERROR 42000: Variable 'optimizer_switch' can't be set to the value of '4' set optimizer_switch=NULL; @@ -1458,21 +1458,21 @@ set optimizer_switch='index_merge=off,index_merge_union=off,default'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set @@global.optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on # # Check index_merge's @@optimizer_switch flags # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, c int, filler char(100), @@ -1582,5 +1582,5 @@ set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on drop table t0, t1; === modified file 'mysql-test/r/innodb_mysql.result' --- a/mysql-test/r/innodb_mysql.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mysql.result 2010-03-11 21:43:31 +0000 @@ -1425,12 +1425,15 @@ # create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables 2 DEPENDENT SUBQUERY t1 system NULL NULL NULL NULL 0 const row not found 2 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 1 +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; End of 5.0 tests CREATE TABLE `t2` ( === modified file 'mysql-test/r/myisam_mrr.result' --- a/mysql-test/r/myisam_mrr.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/myisam_mrr.result 2010-03-11 21:43:31 +0000 @@ -394,7 +394,7 @@ # - engine_condition_pushdown does not affect ICP select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, key(a)); === modified file 'mysql-test/r/ps.result' --- a/mysql-test/r/ps.result 2009-05-27 15:19:44 +0000 +++ b/mysql-test/r/ps.result 2010-03-11 21:43:31 +0000 @@ -149,6 +149,8 @@ c32 set('monday', 'tuesday', 'wednesday') ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -177,6 +179,7 @@ 2 DEPENDENT SUBQUERY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; set @arg00=1; prepare stmt1 from ' create table t1 (m int) as select 1 as m ' ; execute stmt1 ; === modified file 'mysql-test/r/subselect.result' --- a/mysql-test/r/subselect.result 2010-02-17 21:59:41 +0000 +++ b/mysql-test/r/subselect.result 2010-03-11 21:43:31 +0000 @@ -1,4 +1,6 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4803,4 +4805,5 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. === modified file 'mysql-test/r/subselect3.result' --- a/mysql-test/r/subselect3.result 2010-02-17 10:05:27 +0000 +++ b/mysql-test/r/subselect3.result 2010-03-11 21:43:31 +0000 @@ -63,12 +63,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -692,6 +695,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -759,6 +764,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -960,7 +966,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -977,7 +983,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect3_jcl6.result' --- a/mysql-test/r/subselect3_jcl6.result 2010-02-17 10:47:55 +0000 +++ b/mysql-test/r/subselect3_jcl6.result 2010-03-11 21:43:31 +0000 @@ -67,12 +67,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -696,6 +699,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -763,6 +768,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -964,7 +970,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -981,7 +987,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect_no_mat.result' --- a/mysql-test/r/subselect_no_mat.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_mat.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_opts.result' --- a/mysql-test/r/subselect_no_opts.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_opts.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off,semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_semijoin.result' --- a/mysql-test/r/subselect_no_semijoin.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_semijoin.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-15 15:09:35 +0000 @@ -202,39 +202,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 06:32:54 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 15:09:35 +0000 @@ -206,39 +206,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/t/ps.test' --- a/mysql-test/t/ps.test 2009-05-27 15:19:44 +0000 +++ b/mysql-test/t/ps.test 2010-03-11 21:43:31 +0000 @@ -163,6 +163,9 @@ ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -171,6 +174,8 @@ deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # parameters from variables (for field creation) # === modified file 'mysql-test/t/subselect.test' --- a/mysql-test/t/subselect.test 2010-01-17 20:52:20 +0000 +++ b/mysql-test/t/subselect.test 2010-03-11 21:43:31 +0000 @@ -11,6 +11,9 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; --enable_warnings +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + select (select 2); explain extended select (select 2); SELECT (SELECT 1) UNION SELECT (SELECT 2); @@ -4061,4 +4064,6 @@ (SELECT LAST_INSERT_ID() FROM t1 ORDER BY MIN(a) ASC LIMIT 1); DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; + --echo End of 5.1 tests. === modified file 'mysql-test/t/subselect3.test' --- a/mysql-test/t/subselect3.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect3.test 2010-03-11 21:43:31 +0000 @@ -59,9 +59,13 @@ show status like 'Handler_read_rnd_next'; select ' ^ This must show 11' Z; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + # This must show trigcond: explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; # @@ -529,6 +533,9 @@ DROP TABLE t1, t2; +# The next three test cases must be executed with the IN=>EXISTS strategy +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; # # Bug #27870: crash of an equijoin query with WHERE condition containing @@ -588,6 +595,8 @@ DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #34763: item_subselect.cc:1235:Item_in_subselect::row_value_transformer: # Assertion failed, unexpected error message: === modified file 'sql/item_cmpfunc.h' --- a/sql/item_cmpfunc.h 2010-03-13 20:04:52 +0000 +++ b/sql/item_cmpfunc.h 2010-03-15 14:34:56 +0000 @@ -350,6 +350,7 @@ CHARSET_INFO *compare_collation() { return cmp.cmp_collation.collation; } uint decimal_precision() const { return 1; } void top_level_item() { abort_on_null= TRUE; } + Arg_comparator *get_comparator() { return &cmp; } friend class Arg_comparator; }; === modified file 'sql/item_subselect.cc' --- a/sql/item_subselect.cc 2010-02-21 06:32:23 +0000 +++ b/sql/item_subselect.cc 2010-03-09 10:14:06 +0000 @@ -138,6 +138,7 @@ left_expr_cache= NULL; } first_execution= TRUE; + is_constant= FALSE; Item_subselect::cleanup(); DBUG_VOID_RETURN; } @@ -449,8 +450,10 @@ int res; if (thd->is_error()) - /* Do not execute subselect in case of a fatal error */ + { + /* Do not execute subselect in case of a fatal error */ return 1; + } /* Simulate a failure in sub-query execution. Used to test e.g. out of memory or query being killed conditions. @@ -475,9 +478,6 @@ bool Item_in_subselect::exec() { DBUG_ENTER("Item_in_subselect::exec"); - DBUG_ASSERT(exec_method != MATERIALIZATION || - (exec_method == MATERIALIZATION && - engine->engine_type() == subselect_engine::HASH_SJ_ENGINE)); /* Initialize the cache of the left predicate operand. This has to be done as late as now, because Cached_item directly contains a resolved field (not @@ -493,14 +493,14 @@ if (!left_expr_cache && exec_method == MATERIALIZATION) init_left_expr_cache(); - /* If the new left operand is already in the cache, reuse the old result. */ - if (left_expr_cache && test_if_item_cache_changed(*left_expr_cache) < 0) - { - /* Always compute IN for the first row as the cache is not valid for it. */ - if (!first_execution) - DBUG_RETURN(FALSE); - first_execution= FALSE; - } + /* + If the new left operand is already in the cache, reuse the old result. + Use the cached result only if this is not the first execution of IN + because the cache is not valid for the first execution. + */ + if (!first_execution && left_expr_cache && + test_if_item_cache_changed(*left_expr_cache) < 0) + DBUG_RETURN(FALSE); /* The exec() method below updates item::value, and item::null_value, thus if @@ -910,8 +910,8 @@ Item_in_subselect::Item_in_subselect(Item * left_exp, st_select_lex *select_lex): Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), - upper_item(0) + is_constant(FALSE), optimizer(0), pushed_cond_guards(NULL), + exec_method(NOT_TRANSFORMED), upper_item(0) { DBUG_ENTER("Item_in_subselect::Item_in_subselect"); left_expr= left_exp; @@ -1105,6 +1105,8 @@ { DBUG_ASSERT(fixed == 1); null_value= 0; + if (is_constant) + return value; if (exec()) { reset(); @@ -1571,9 +1573,9 @@ DBUG_ENTER("Item_in_subselect::row_value_transformer"); // psergey: duplicated_subselect_card_check - if (select_lex->item_list.elements != left_expr->cols()) + if (select_lex->item_list.elements != cols_num) { - my_error(ER_OPERAND_COLUMNS, MYF(0), left_expr->cols()); + my_error(ER_OPERAND_COLUMNS, MYF(0), cols_num); DBUG_RETURN(RES_ERROR); } @@ -1980,17 +1982,69 @@ bool Item_in_subselect::fix_fields(THD *thd_arg, Item **ref) { - bool result = 0; + uint outer_cols_num; + List<Item> *inner_cols; if (exec_method == SEMI_JOIN) return !( (*ref)= new Item_int(1)); - if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed) - result = left_expr->fix_fields(thd_arg, &left_expr); - - return result || Item_subselect::fix_fields(thd_arg, ref); + /* + Check if the outer and inner IN operands match in those cases when we + will not perform IN=>EXISTS transformation. Currently this is when we + use subquery materialization. + + The condition below is true when this method was called recursively from + inside JOIN::prepare for the JOIN object created by the call chain + Item_subselect::fix_fields -> subselect_single_select_engine::prepare, + which creates a JOIN object for the subquery and calls JOIN::prepare for + the JOIN of the subquery. + Notice that in some cases, this doesn't happen, and the check_cols() + test for each Item happens later in + Item_in_subselect::row_value_in_to_exists_transformer. + The reason for this mess is that our JOIN::prepare phase works top-down + instead of bottom-up, so we first do name resoluton and semantic checks + for the outer selects, then for the inner. + */ + if (engine && + engine->engine_type() == subselect_engine::SINGLE_SELECT_ENGINE && + ((subselect_single_select_engine*)engine)->join) + { + outer_cols_num= left_expr->cols(); + + if (unit->is_union()) + inner_cols= &(unit->types); + else + inner_cols= &(unit->first_select()->item_list); + if (outer_cols_num != inner_cols->elements) + { + my_error(ER_OPERAND_COLUMNS, MYF(0), outer_cols_num); + return TRUE; + } + if (outer_cols_num > 1) + { + List_iterator<Item> inner_col_it(*inner_cols); + Item *inner_col; + for (uint i= 0; i < outer_cols_num; i++) + { + inner_col= inner_col_it++; + if (inner_col->check_cols(left_expr->element_index(i)->cols())) + return TRUE; + } + } + } + + if (thd_arg->lex->view_prepare_mode && left_expr && !left_expr->fixed && + left_expr->fix_fields(thd_arg, &left_expr)) + return TRUE; + if (Item_subselect::fix_fields(thd_arg, ref)) + return TRUE; + + fixed= TRUE; + + return FALSE; } + void Item_in_subselect::fix_after_pullout(st_select_lex *new_parent, Item **ref) { left_expr->fix_after_pullout(new_parent, &left_expr); @@ -2267,10 +2321,9 @@ void subselect_uniquesubquery_engine::cleanup() { DBUG_ENTER("subselect_uniquesubquery_engine::cleanup"); - /* - subselect_uniquesubquery_engine have not 'result' assigbed, so we do not - cleanup() it - */ + /* Tell handler we don't need the index anymore */ + if (tab->table->file->inited) + tab->table->file->ha_index_end(); DBUG_VOID_RETURN; } @@ -2291,7 +2344,7 @@ Create and prepare the JOIN object that represents the query execution plan for the subquery. - @detail + @details This method is called from Item_subselect::fix_fields. For prepared statements it is called both during the PREPARE and EXECUTE phases in the following ways: @@ -2593,14 +2646,23 @@ for (;;) { error=table->file->ha_rnd_next(table->record[0]); - if (error && error != HA_ERR_END_OF_FILE) - { - error= report_error(table, error); - break; + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(table, error); + break; + } } - /* No more rows */ - if (table->status) - break; if (!cond || cond->val_int()) { @@ -2711,6 +2773,56 @@ /* + @retval 1 A NULL was found in the outer reference, index lookup is + not applicable, the outer ref is unsusable as a lookup key, + use some other method to find a match. + @retval 0 The outer ref was copied into an index lookup key. + @retval -1 The outer ref cannot possibly match any row, IN is FALSE. +*/ +/* TIMOUR: this method is a variant of copy_ref_key(), needs refactoring. */ + +int subselect_uniquesubquery_engine::copy_ref_key_simple() +{ + for (store_key **copy= tab->ref.key_copy ; *copy ; copy++) + { + enum store_key::store_key_result store_res; + store_res= (*copy)->copy(); + tab->ref.key_err= store_res; + + /* + When there is a NULL part in the key we don't need to make index + lookup for such key thus we don't need to copy whole key. + If we later should do a sequential scan return OK. Fail otherwise. + + See also the comment for the subselect_uniquesubquery_engine::exec() + function. + */ + null_keypart= (*copy)->null_key; + if (null_keypart) + return 1; + + /* + Check if the error is equal to STORE_KEY_FATAL. This is not expressed + using the store_key::store_key_result enum because ref.key_err is a + boolean and we want to detect both TRUE and STORE_KEY_FATAL from the + space of the union of the values of [TRUE, FALSE] and + store_key::store_key_result. + TODO: fix the variable an return types. + */ + if (store_res == store_key::STORE_KEY_FATAL) + { + /* + Error converting the left IN operand to the column type of the right + IN operand. + */ + return -1; + } + } + return 0; +} + + +/* Execute subselect SYNOPSIS @@ -2750,7 +2862,13 @@ /* TODO: change to use of 'full_scan' here? */ if (copy_ref_key()) + { + /* + TIMOUR: copy_ref_key() == 1 means NULL result, not error, why return 1? + Check who reiles on this result. + */ DBUG_RETURN(1); + } if (table->status) { /* @@ -2791,6 +2909,46 @@ } +/* + TIMOUR: write comment +*/ + +int subselect_uniquesubquery_engine::index_lookup() +{ + DBUG_ENTER("subselect_uniquesubquery_engine::index_lookup"); + int error; + TABLE *table= tab->table; + + if (!table->file->inited) + table->file->ha_index_init(tab->ref.key, 0); + error= table->file->ha_index_read_map(table->record[0], + tab->ref.key_buff, + make_prev_keypart_map(tab-> + ref.key_parts), + HA_READ_KEY_EXACT); + DBUG_PRINT("info", ("lookup result: %i", error)); + + if (error && error != HA_ERR_KEY_NOT_FOUND && error != HA_ERR_END_OF_FILE) + { + /* + TIMOUR: I don't understand at all when do we need to call report_error. + In most places where we access an index, we don't do this. Why here? + */ + error= report_error(table, error); + DBUG_RETURN(error); + } + + table->null_row= 0; + if (!error && (!cond || cond->val_int())) + ((Item_in_subselect *) item)->value= 1; + else + ((Item_in_subselect *) item)->value= 0; + + DBUG_RETURN(0); +} + + + subselect_uniquesubquery_engine::~subselect_uniquesubquery_engine() { /* Tell handler we don't need the index anymore */ @@ -3225,6 +3383,7 @@ bool subselect_uniquesubquery_engine::no_tables() { /* returning value is correct, but this method should never be called */ + DBUG_ASSERT(FALSE); return 0; } @@ -3235,16 +3394,259 @@ /** + Check if an IN predicate should be executed via partial matching using + only schema information. + + @details + This test essentially has three results: + - partial matching is applicable, but cannot be executed due to a + limitation in the total number of indexes, as a result we can't + use subquery materialization at all. + - partial matching is either applicable or not, and this can be + determined by looking at 'this->max_keys'. + If max_keys > 1, then we need partial matching because there are + more indexes than just the one we use during materialization to + remove duplicates. + + @note + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_schema() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + + if (item_in->is_top_level_item()) + return COMPLETE_MATCH; + else + { + List_iterator<Item> inner_col_it(*item_in->unit->get_unit_column_types()); + Item *outer_col, *inner_col; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + outer_col= item_in->left_expr->element_index(i); + inner_col= inner_col_it++; + + if (!inner_col->maybe_null && !outer_col->maybe_null) + bitmap_set_bit(&non_null_key_parts, i); + else + { + bitmap_set_bit(&partial_match_key_parts, i); + ++count_partial_match_columns; + } + } + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (count_partial_match_columns) + return PARTIAL_MATCH; + return COMPLETE_MATCH; +} + + +/** + Test whether an IN predicate must be computed via partial matching + based on the NULL statistics for each column of a materialized subquery. + + @details The procedure analyzes column NULL statistics, updates the + matching type of columns that cannot be NULL or that contain only NULLs. + Based on this, the procedure determines the final execution strategy for + the [NOT] IN predicate. + + @retval PARTIAL_MATCH if a partial match should be used + @retval COMPLETE_MATCH if a complete match (index lookup) should be used +*/ + +subselect_hash_sj_engine::exec_strategy +subselect_hash_sj_engine::get_strategy_using_data() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + Item *outer_col; + + /* + If we already determined that a complete match is enough based on schema + information, nothing can be better. + */ + if (strategy == COMPLETE_MATCH) + return COMPLETE_MATCH; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + if (!bitmap_is_set(&partial_match_key_parts, i)) + continue; + outer_col= item_in->left_expr->element_index(i); + /* + If column 'i' doesn't contain NULLs, and the corresponding outer reference + cannot have a NULL value, then 'i' is a non-nullable column. + */ + if (result_sink->get_null_count_of_col(i) == 0 && !outer_col->maybe_null) + { + bitmap_clear_bit(&partial_match_key_parts, i); + bitmap_set_bit(&non_null_key_parts, i); + --count_partial_match_columns; + } + if (result_sink->get_null_count_of_col(i) == + tmp_table->file->stats.records) + ++count_null_only_columns; + } + + /* If no column contains NULLs use regular hash index lookups. */ + if (!count_partial_match_columns) + return COMPLETE_MATCH; + return PARTIAL_MATCH; +} + + +void +subselect_hash_sj_engine::choose_partial_match_strategy( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t pm_buff_size; + + DBUG_ASSERT(strategy == PARTIAL_MATCH); + /* + Choose according to global optimizer switch. If only one of the switches is + 'ON', then the remaining strategy is the only possible one. The only cases + when this will be overriden is when the total size of all buffers for the + merge strategy is bigger than the 'rowid_merge_buff_size' system variable, + or if there isn't enough physical memory to allocate the buffers. + */ + if (!optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_SCAN; + else if + ( optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) && + !optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) + strategy= PARTIAL_MATCH_MERGE; + + /* + If both switches are ON, or both are OFF, we interpret that as "let the + optimizer decide". Perform a cost based choice between the two partial + matching strategies. + */ + /* + TIMOUR: the above interpretation of the switch values could be changed to: + - if both are ON - let the optimizer decide, + - if both are OFF - do not use partial matching, therefore do not use + materialization in non-top-level predicates. + The problem with this is that we know for sure if we need partial matching + only after the subquery is materialized, and this is too late to revert to + the IN=>EXISTS strategy. + */ + if (strategy == PARTIAL_MATCH) + { + /* + TIMOUR: Currently we use a super simplistic measure. This will be + addressed in a separate task. + */ + if (tmp_table->file->stats.records < 100) + strategy= PARTIAL_MATCH_SCAN; + else + strategy= PARTIAL_MATCH_MERGE; + } + + /* Check if there is enough memory for the rowid merge strategy. */ + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_buff_size= rowid_merge_buff_size(has_non_null_key, + has_covering_null_row, + partial_match_key_parts); + if (pm_buff_size > thd->variables.rowid_merge_buff_size) + strategy= PARTIAL_MATCH_SCAN; + } +} + + +/* + Compute the memory size of all buffers proportional to the number of rows + in tmp_table. + + @details + If the result is bigger than thd->variables.rowid_merge_buff_size, partial + matching via merging is not applicable. +*/ + +size_t subselect_hash_sj_engine::rowid_merge_buff_size( + bool has_non_null_key, bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts) +{ + size_t buff_size; /* Total size of all buffers used by partial matching. */ + ha_rows row_count= tmp_table->file->stats.records; + uint rowid_length= tmp_table->file->ref_length; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + /* Size of the subselect_rowid_merge_engine::row_num_to_rowid buffer. */ + buff_size= row_count * rowid_length * sizeof(uchar); + + if (has_non_null_key) + { + /* Add the size of Ordered_key::key_buff of the only non-NULL key. */ + buff_size+= row_count * sizeof(rownum_t); + } + + if (!has_covering_null_row) + { + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i) || + result_sink->get_null_count_of_col(i) == row_count) + continue; /* In these cases we wouldn't construct Ordered keys. */ + + /* Add the size of Ordered_key::key_buff */ + buff_size+= (row_count - result_sink->get_null_count_of_col(i)) * + sizeof(rownum_t); + /* Add the size of Ordered_key::null_key */ + buff_size+= bitmap_buffer_size(result_sink->get_max_null_of_col(i)); + } + } + + return buff_size; +} + + +/* + Initialize a MY_BITMAP with a buffer allocated on the current + memory root. + TIMOUR: move to bitmap C file? +*/ + +static my_bool +bitmap_init_memroot(MY_BITMAP *map, uint n_bits, MEM_ROOT *mem_root) +{ + my_bitmap_map *bitmap_buf; + + if (!(bitmap_buf= (my_bitmap_map*) alloc_root(mem_root, + bitmap_buffer_size(n_bits))) || + bitmap_init(map, bitmap_buf, n_bits, FALSE)) + return TRUE; + bitmap_clear_all(map); + return FALSE; +} + + +/** Create all structures needed for IN execution that can live between PS reexecution. - @detail + @param tmp_columns the items that produce the data for the temp table + + @details - Create a temporary table to store the result of the IN subquery. The temporary table has one hash index on all its columns. - Create a new result sink that sends the result stream of the subquery to the temporary table, - - Create and initialize a new JOIN_TAB, and TABLE_REF objects to perform - lookups into the indexed temporary table. @notice: Currently Item_subselect::init() already chooses and creates at parse @@ -3256,71 +3658,178 @@ bool subselect_hash_sj_engine::init_permanent(List<Item> *tmp_columns) { - /* The result sink where we will materialize the subquery result. */ - select_union *tmp_result_sink; - /* The table into which the subquery is materialized. */ - TABLE *tmp_table; - KEY *tmp_key; /* The only index on the temporary table. */ - uint tmp_key_parts; /* Number of keyparts in tmp_key. */ - Item_in_subselect *item_in= (Item_in_subselect *) item; + /* Options to create_tmp_table. */ + ulonglong tmp_create_options= thd->options | TMP_TABLE_ALL_COLUMNS; + /* | TMP_TABLE_FORCE_MYISAM; TIMOUR: force MYISAM */ DBUG_ENTER("subselect_hash_sj_engine::init_permanent"); - /* 1. Create/initialize materialization related objects. */ + if (bitmap_init_memroot(&non_null_key_parts, tmp_columns->elements, + thd->mem_root) || + bitmap_init_memroot(&partial_match_key_parts, tmp_columns->elements, + thd->mem_root)) + DBUG_RETURN(TRUE); /* Create and initialize a select result interceptor that stores the result stream in a temporary table. The temporary table itself is managed (created/filled/etc) internally by the interceptor. */ - if (!(tmp_result_sink= new select_union)) - DBUG_RETURN(TRUE); - if (tmp_result_sink->create_result_table( - thd, tmp_columns, TRUE, - thd->options | TMP_TABLE_ALL_COLUMNS, +/* + TIMOUR: + Select a more efficient result sink when we know there is no need to collect + data statistics. + + if (strategy == COMPLETE_MATCH) + { + if (!(result= new select_union)) + DBUG_RETURN(TRUE); + } + else if (strategy == PARTIAL_MATCH) + { + if (!(result= new select_materialize_with_stats)) + DBUG_RETURN(TRUE); + } +*/ + if (!(result= new select_materialize_with_stats)) + DBUG_RETURN(TRUE); + + if (((select_union*) result)->create_result_table( + thd, tmp_columns, TRUE, tmp_create_options, "materialized subselect", TRUE)) DBUG_RETURN(TRUE); - tmp_table= tmp_result_sink->table; - tmp_key= tmp_table->key_info; - tmp_key_parts= tmp_key->key_parts; + tmp_table= ((select_union*) result)->table; /* - If the subquery has blobs, or the total key lenght is bigger than some - length, then the created index cannot be used for lookups and we - can't use hash semi join. If this is the case, delete the temporary - table since it will not be used, and tell the caller we failed to - initialize the engine. + If the subquery has blobs, or the total key lenght is bigger than + some length, or the total number of key parts is more than the + allowed maximum (currently MAX_REF_PARTS == 16), then the created + index cannot be used for lookups and we can't use hash semi + join. If this is the case, delete the temporary table since it + will not be used, and tell the caller we failed to initialize the + engine. */ if (tmp_table->s->keys == 0) { -#ifndef DBUG_OFF - handlerton *tmp_table_hton= tmp_table->s->db_type(); -#ifdef USE_MARIA_FOR_TMP_TABLES - DBUG_ASSERT(tmp_table_hton == maria_hton); -#else - DBUG_ASSERT(tmp_table_hton == myisam_hton); -#endif -#endif DBUG_ASSERT( tmp_table->s->uniques || tmp_table->key_info->key_length >= tmp_table->file->max_key_length() || tmp_table->key_info->key_parts > tmp_table->file->max_key_parts()); free_tmp_table(thd, tmp_table); + tmp_table= NULL; delete result; result= NULL; DBUG_RETURN(TRUE); } - result= tmp_result_sink; /* Make sure there is only one index on the temp table, and it doesn't have the extra key part created when s->uniques > 0. */ - DBUG_ASSERT(tmp_table->s->keys == 1 && tmp_columns->elements == tmp_key_parts); - - - /* 2. Create/initialize execution related objects. */ + DBUG_ASSERT(tmp_table->s->keys == 1 && + ((Item_in_subselect *) item)->left_expr->cols() == + tmp_table->key_info->key_parts); + + if (make_semi_join_conds() || + /* A unique_engine is used both for complete and partial matching. */ + !(lookup_engine= make_unique_engine())) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} + + +/* + Create an artificial condition to post-filter those rows matched by index + lookups that cannot be distinguished by the index lookup procedure. + + @notes + The need for post-filtering may occur e.g. because of + truncation. Prepared statements execution requires that fix_fields is + called for every execution. In order to call fix_fields we need to + create a Name_resolution_context and a corresponding TABLE_LIST for + the temporary table for the subquery, so that all column references + to the materialized subquery table can be resolved correctly. + + @returns + @retval TRUE memory allocation error occurred + @retval FALSE the conditions were created and resolved (fixed) +*/ + +bool subselect_hash_sj_engine::make_semi_join_conds() +{ + /* + Table reference for tmp_table that is used to resolve column references + (Item_fields) to columns in tmp_table. + */ + TABLE_LIST *tmp_table_ref; + /* Name resolution context for all tmp_table columns created below. */ + Name_resolution_context *context; + Item_in_subselect *item_in= (Item_in_subselect *) item; + + DBUG_ENTER("subselect_hash_sj_engine::make_semi_join_conds"); + DBUG_ASSERT(semi_join_conds == NULL); + + if (!(semi_join_conds= new Item_cond_and)) + DBUG_RETURN(TRUE); + + if (!(tmp_table_ref= (TABLE_LIST*) thd->alloc(sizeof(TABLE_LIST)))) + DBUG_RETURN(TRUE); + + tmp_table_ref->init_one_table("", "materialized subselect", TL_READ); + tmp_table_ref->table= tmp_table; + + context= new Name_resolution_context; + context->init(); + context->first_name_resolution_table= + context->last_name_resolution_table= tmp_table_ref; + + for (uint i= 0; i < item_in->left_expr->cols(); i++) + { + Item_func_eq *eq_cond; /* New equi-join condition for the current column. */ + /* Item for the corresponding field from the materialized temp table. */ + Item_field *right_col_item; + + if (!(right_col_item= new Item_field(thd, context, tmp_table->field[i])) || + !(eq_cond= new Item_func_eq(item_in->left_expr->element_index(i), + right_col_item)) || + (((Item_cond_and*)semi_join_conds)->add(eq_cond))) + { + delete semi_join_conds; + semi_join_conds= NULL; + DBUG_RETURN(TRUE); + } + } + if (semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) + DBUG_RETURN(TRUE); + + DBUG_RETURN(FALSE); +} + + +/** + Create a new uniquesubquery engine for the execution of an IN predicate. + + @details + Create and initialize a new JOIN_TAB, and Table_ref objects to perform + lookups into the indexed temporary table. + + @retval A new subselect_hash_sj_engine object + @retval NULL if a memory allocation error occurs +*/ + +subselect_uniquesubquery_engine* +subselect_hash_sj_engine::make_unique_engine() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + /* The only index on the temporary table. */ + KEY *tmp_key= tmp_table->key_info; + /* Number of keyparts in tmp_key. */ + uint tmp_key_parts= tmp_key->key_parts; + JOIN_TAB *tab; + + DBUG_ENTER("subselect_hash_sj_engine::make_unique_engine"); /* Create and initialize the JOIN_TAB that represents an index lookup @@ -3328,9 +3837,9 @@ - this JOIN_TAB has no corresponding JOIN (and doesn't need one), and - here we initialize only those members that are used by subselect_uniquesubquery_engine, so these objects are incomplete. - */ + */ if (!(tab= (JOIN_TAB*) thd->alloc(sizeof(JOIN_TAB)))) - DBUG_RETURN(TRUE); + DBUG_RETURN(NULL); tab->table= tmp_table; tab->ref.key= 0; /* The only temp table index. */ tab->ref.key_length= tmp_key->key_length; @@ -3341,60 +3850,18 @@ (tmp_key_parts + 1)))) || !(tab->ref.items= (Item**) thd->alloc(sizeof(Item*) * tmp_key_parts))) - DBUG_RETURN(TRUE); + DBUG_RETURN(NULL); KEY_PART_INFO *cur_key_part= tmp_key->key_part; store_key **ref_key= tab->ref.key_copy; uchar *cur_ref_buff= tab->ref.key_buff; - - /* - Create an artificial condition to post-filter those rows matched by index - lookups that cannot be distinguished by the index lookup procedure, e.g. - because of truncation. Prepared statements execution requires that - fix_fields is called for every execution. In order to call fix_fields we - need to create a Name_resolution_context and a corresponding TABLE_LIST - for the temporary table for the subquery, so that all column references - to the materialized subquery table can be resolved correctly. - */ - DBUG_ASSERT(cond == NULL); - if (!(cond= new Item_cond_and)) - DBUG_RETURN(TRUE); - /* - Table reference for tmp_table that is used to resolve column references - (Item_fields) to columns in tmp_table. - */ - TABLE_LIST *tmp_table_ref; - if (!(tmp_table_ref= (TABLE_LIST*) thd->alloc(sizeof(TABLE_LIST)))) - DBUG_RETURN(TRUE); - - tmp_table_ref->init_one_table("", "materialized subselect", TL_READ); - tmp_table_ref->table= tmp_table; - - /* Name resolution context for all tmp_table columns created below. */ - Name_resolution_context *context= new Name_resolution_context; - context->init(); - context->first_name_resolution_table= - context->last_name_resolution_table= tmp_table_ref; for (uint i= 0; i < tmp_key_parts; i++, cur_key_part++, ref_key++) { - Item_func_eq *eq_cond; /* New equi-join condition for the current column. */ - /* Item for the corresponding field from the materialized temp table. */ - Item_field *right_col_item; + tab->ref.items[i]= item_in->left_expr->element_index(i); int null_count= test(cur_key_part->field->real_maybe_null()); - tab->ref.items[i]= item_in->left_expr->element_index(i); - - if (!(right_col_item= new Item_field(thd, context, cur_key_part->field)) || - !(eq_cond= new Item_func_eq(tab->ref.items[i], right_col_item)) || - ((Item_cond_and*)cond)->add(eq_cond)) - { - delete cond; - cond= NULL; - DBUG_RETURN(TRUE); - } - *ref_key= new store_key_item(thd, cur_key_part->field, - /* TODO: + /* TIMOUR: the NULL byte is taken into account in cur_key_part->store_length, so instead of cur_ref_buff + test(maybe_null), we could @@ -3409,10 +3876,8 @@ tab->ref.key_err= 1; tab->ref.key_parts= tmp_key_parts; - if (cond->fix_fields(thd, &cond)) - DBUG_RETURN(TRUE); - - DBUG_RETURN(FALSE); + DBUG_RETURN(new subselect_uniquesubquery_engine(thd, tab, item, + semi_join_conds)); } @@ -3435,7 +3900,8 @@ Repeat name resolution for 'cond' since cond is not part of any clause of the query, and it is not 'fixed' during JOIN::prepare. */ - if (cond && !cond->fixed && cond->fix_fields(thd, &cond)) + if (semi_join_conds && !semi_join_conds->fixed && + semi_join_conds->fix_fields(thd, (Item**)&semi_join_conds)) return TRUE; /* Let our engine reuse this query plan for materialization. */ materialize_join= materialize_engine->join; @@ -3446,32 +3912,53 @@ subselect_hash_sj_engine::~subselect_hash_sj_engine() { + delete lookup_engine; delete result; - if (tab) - free_tmp_table(thd, tab->table); + if (tmp_table) + free_tmp_table(thd, tmp_table); } /** Cleanup performed after each PS execution. - @detail + @details Called in the end of JOIN::prepare for PS from Item_subselect::cleanup. */ void subselect_hash_sj_engine::cleanup() { + enum_engine_type lookup_engine_type= lookup_engine->engine_type(); is_materialized= FALSE; + bitmap_clear_all(&non_null_key_parts); + bitmap_clear_all(&partial_match_key_parts); + count_partial_match_columns= 0; + count_null_only_columns= 0; + strategy= UNDEFINED; + materialize_engine->cleanup(); + if (lookup_engine_type == TABLE_SCAN_ENGINE || + lookup_engine_type == ROWID_MERGE_ENGINE) + { + subselect_engine *inner_lookup_engine; + inner_lookup_engine= + ((subselect_partial_match_engine*) lookup_engine)->lookup_engine; + /* + Partial match engines are recreated for each PS execution inside + subselect_hash_sj_engine::exec(). + */ + delete lookup_engine; + lookup_engine= inner_lookup_engine; + } + DBUG_ASSERT(lookup_engine->engine_type() == UNIQUESUBQUERY_ENGINE); + lookup_engine->cleanup(); result->cleanup(); /* Resets the temp table as well. */ - materialize_engine->cleanup(); - subselect_uniquesubquery_engine::cleanup(); } /** Execute a subquery IN predicate via materialization. - @detail + @details If needed materialize the subquery into a temporary table, then copmpute the predicate via a lookup into this table. @@ -3482,6 +3969,9 @@ int subselect_hash_sj_engine::exec() { Item_in_subselect *item_in= (Item_in_subselect *) item; + SELECT_LEX *save_select= thd->lex->current_select; + subselect_partial_match_engine *pm_engine= NULL; + int res= 0; DBUG_ENTER("subselect_hash_sj_engine::exec"); @@ -3489,56 +3979,126 @@ Optimize and materialize the subquery during the first execution of the subquery predicate. */ - if (!is_materialized) - { - int res= 0; - SELECT_LEX *save_select= thd->lex->current_select; - thd->lex->current_select= materialize_engine->select_lex; - if ((res= materialize_join->optimize())) - goto err; /* purecov: inspected */ - materialize_join->exec(); - if ((res= test(materialize_join->error || thd->is_fatal_error))) - goto err; - - /* - TODO: - - Unlock all subquery tables as we don't need them. To implement this - we need to add new functionality to JOIN::join_free that can unlock - all tables in a subquery (and all its subqueries). - - The temp table used for grouping in the subquery can be freed - immediately after materialization (yet it's done together with - unlocking). - */ - is_materialized= TRUE; - /* - If the subquery returned no rows, the temporary table is empty, so we know - directly that the result of IN is FALSE. We first update the table - statistics, then we test if the temporary table for the query result is - empty. - */ - tab->table->file->info(HA_STATUS_VARIABLE); - if (!tab->table->file->stats.records) - { - empty_result_set= TRUE; - item_in->value= FALSE; - /* TODO: check we need this: item_in->null_value= FALSE; */ - DBUG_RETURN(FALSE); - } - /* Set tmp_param only if its usable, i.e. tmp_param->copy_field != NULL. */ - tmp_param= &(item_in->unit->outer_select()->join->tmp_table_param); - if (tmp_param && !tmp_param->copy_field) - tmp_param= NULL; + thd->lex->current_select= materialize_engine->select_lex; + if ((res= materialize_join->optimize())) + goto err; /* purecov: inspected */ + DBUG_ASSERT(!is_materialized); /* We should materialize only once. */ + materialize_join->exec(); + if ((res= test(materialize_join->error || thd->is_fatal_error))) + goto err; + + /* + TODO: + - Unlock all subquery tables as we don't need them. To implement this + we need to add new functionality to JOIN::join_free that can unlock + all tables in a subquery (and all its subqueries). + - The temp table used for grouping in the subquery can be freed + immediately after materialization (yet it's done together with + unlocking). + */ + is_materialized= TRUE; + /* + If the subquery returned no rows, the temporary table is empty, so we know + directly that the result of IN is FALSE. We first update the table + statistics, then we test if the temporary table for the query result is + empty. + */ + tmp_table->file->info(HA_STATUS_VARIABLE); + if (!tmp_table->file->stats.records) + { + item_in->value= FALSE; + /* The value of IN will not change during this execution. */ + item_in->is_constant= TRUE; + item_in->set_first_execution(); + /* TIMOUR: check if we need this: item_in->null_value= FALSE; */ + DBUG_RETURN(FALSE); + } + + /* + TIMOUR: The schema-based analysis for partial matching can be done once for + prepared statement and remembered. It is done here to remove the need to + save/restore all related variables between each re-execution, thus making + the code simpler. + */ + strategy= get_strategy_using_schema(); + /* This call may discover that we don't need partial matching at all. */ + strategy= get_strategy_using_data(); + if (strategy == PARTIAL_MATCH) + { + uint count_pm_keys; /* Total number of keys needed for partial matching. */ + MY_BITMAP *nn_key_parts; /* The key parts of the only non-NULL index. */ + uint covering_null_row_width; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + + nn_key_parts= (count_partial_match_columns < tmp_table->s->fields) ? + &non_null_key_parts : NULL; + + if (result_sink->get_max_nulls_in_row() == + tmp_table->s->fields - + (nn_key_parts ? bitmap_bits_set(nn_key_parts) : 0)) + covering_null_row_width= result_sink->get_max_nulls_in_row(); + else + covering_null_row_width= 0; + + if (covering_null_row_width) + count_pm_keys= nn_key_parts ? 1 : 0; + else + count_pm_keys= count_partial_match_columns - count_null_only_columns + + (nn_key_parts ? 1 : 0); + + choose_partial_match_strategy(test(nn_key_parts), + test(covering_null_row_width), + &partial_match_key_parts); + DBUG_ASSERT(strategy == PARTIAL_MATCH_MERGE || + strategy == PARTIAL_MATCH_SCAN); + if (strategy == PARTIAL_MATCH_MERGE) + { + pm_engine= + new subselect_rowid_merge_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + count_pm_keys, + covering_null_row_width, + item, result, + semi_join_conds->argument_list()); + if (!pm_engine || + ((subselect_rowid_merge_engine*) pm_engine)-> + init(nn_key_parts, &partial_match_key_parts)) + { + /* + The call to init() would fail if there was not enough memory to allocate + all buffers for the rowid merge strategy. In this case revert to table + scanning which doesn't need any big buffers. + */ + delete pm_engine; + pm_engine= NULL; + strategy= PARTIAL_MATCH_SCAN; + } + } + + if (strategy == PARTIAL_MATCH_SCAN) + { + if (!(pm_engine= + new subselect_table_scan_engine((subselect_uniquesubquery_engine*) + lookup_engine, tmp_table, + item, result, + semi_join_conds->argument_list(), + covering_null_row_width))) + { + /* This is an irrecoverable error. */ + res= 1; + goto err; + } + } + } + + if (pm_engine) + lookup_engine= pm_engine; + item_in->change_engine(lookup_engine); err: - thd->lex->current_select= save_select; - if (res) - DBUG_RETURN(res); - } - - /* - Lookup the left IN operand in the hash index of the materialized subquery. - */ - DBUG_RETURN(subselect_uniquesubquery_engine::exec()); + thd->lex->current_select= save_select; + DBUG_RETURN(res); } @@ -3551,10 +4111,1008 @@ str->append(STRING_WITH_LEN(" <materialize> (")); materialize_engine->print(str, query_type); str->append(STRING_WITH_LEN(" ), ")); - if (tab) - subselect_uniquesubquery_engine::print(str, query_type); + + if (lookup_engine) + lookup_engine->print(str, query_type); else str->append(STRING_WITH_LEN( - "<the access method for lookups is not yet created>" + "<engine selected at execution time>" )); } + +void subselect_hash_sj_engine::fix_length_and_dec(Item_cache** row) +{ + DBUG_ASSERT(FALSE); +} + +void subselect_hash_sj_engine::exclude() +{ + DBUG_ASSERT(FALSE); +} + +bool subselect_hash_sj_engine::no_tables() +{ + DBUG_ASSERT(FALSE); + return FALSE; +} + +bool subselect_hash_sj_engine::change_result(Item_subselect *si, + select_result_interceptor *res) +{ + DBUG_ASSERT(FALSE); + return TRUE; +} + + +Ordered_key::Ordered_key(uint keyid_arg, TABLE *tbl_arg, Item *search_key_arg, + ha_rows null_count_arg, ha_rows min_null_row_arg, + ha_rows max_null_row_arg, uchar *row_num_to_rowid_arg) + : keyid(keyid_arg), tbl(tbl_arg), search_key(search_key_arg), + row_num_to_rowid(row_num_to_rowid_arg), null_count(null_count_arg) +{ + DBUG_ASSERT(tbl->file->stats.records > null_count); + key_buff_elements= tbl->file->stats.records - null_count; + cur_key_idx= HA_POS_ERROR; + + DBUG_ASSERT((null_count && min_null_row_arg && max_null_row_arg) || + (!null_count && !min_null_row_arg && !max_null_row_arg)); + if (null_count) + { + /* The counters are 1-based, for key access we need 0-based indexes. */ + min_null_row= min_null_row_arg - 1; + max_null_row= max_null_row_arg - 1; + } + else + min_null_row= max_null_row= 0; +} + + +Ordered_key::~Ordered_key() +{ + my_free((char*) key_buff, MYF(0)); + bitmap_free(&null_key); +} + + +/* + Cleanup that needs to be done for each PS (re)execution. +*/ + +void Ordered_key::cleanup() +{ + /* + Currently these keys are recreated for each PS re-execution, thus + there is nothing to cleanup, the whole object goes away after execution + is over. All handler related initialization/deinitialization is done by + the parent subselect_rowid_merge_engine object. + */ +} + + +/* + Initialize a multi-column index. +*/ + +bool Ordered_key::init(MY_BITMAP *columns_to_index) +{ + THD *thd= tbl->in_use; + uint cur_key_col= 0; + Item_field *cur_tmp_field; + Item_func_lt *fn_less_than; + + key_column_count= bitmap_bits_set(columns_to_index); + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(key_column_count * + sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(key_column_count * + sizeof(Item_func_lt*)); + + for (uint i= 0; i < columns_to_index->n_bits; i++) + { + if (!bitmap_is_set(columns_to_index, i)) + continue; + cur_tmp_field= new Item_field(tbl->field[i]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + fn_less_than= new Item_func_lt(cur_tmp_field, + search_key->element_index(i)); + fn_less_than->fix_fields(thd, (Item**) &fn_less_than); + key_columns[cur_key_col]= cur_tmp_field; + compare_pred[cur_key_col]= fn_less_than; + ++cur_key_col; + } + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Initialize a single-column index. +*/ + +bool Ordered_key::init(int col_idx) +{ + THD *thd= tbl->in_use; + + key_column_count= 1; + + // TIMOUR: check for mem allocation err, revert to scan + + key_columns= (Item_field**) thd->alloc(sizeof(Item_field*)); + compare_pred= (Item_func_lt**) thd->alloc(sizeof(Item_func_lt*)); + + key_columns[0]= new Item_field(tbl->field[col_idx]); + /* Create the predicate (tmp_column[i] < outer_ref[i]). */ + compare_pred[0]= new Item_func_lt(key_columns[0], + search_key->element_index(col_idx)); + compare_pred[0]->fix_fields(thd, (Item**)&compare_pred[0]); + + if (alloc_keys_buffers()) + { + /* TIMOUR revert to partial match via table scan. */ + return TRUE; + } + return FALSE; +} + + +/* + Allocate the buffers for both the row number, and the NULL-bitmap indexes. +*/ + +bool Ordered_key::alloc_keys_buffers() +{ + DBUG_ASSERT(key_buff_elements > 0); + + if (!(key_buff= (rownum_t*) my_malloc(key_buff_elements * sizeof(rownum_t), + MYF(MY_WME)))) + return TRUE; + + /* + TIMOUR: it is enough to create bitmaps with size + (max_null_row - min_null_row), and then use min_null_row as + lookup offset. + */ + /* Notice that max_null_row is max array index, we need count, so +1. */ + if (bitmap_init(&null_key, NULL, max_null_row + 1, FALSE)) + return TRUE; + + cur_key_idx= HA_POS_ERROR; + + return FALSE; +} + + +/* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + + @retval -1 + @retval 0 + @retval +1 +*/ + +int +Ordered_key::cmp_keys_by_row_data(ha_rows a, ha_rows b) +{ + uchar *rowid_a, *rowid_b; + int error, cmp_res; + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + + if (a == b) + return 0; + /* Get the corresponding rowids. */ + rowid_a= row_num_to_rowid + a * rowid_length; + rowid_b= row_num_to_rowid + b * rowid_length; + /* Fetch the rows for comparison. */ + error= tbl->file->ha_rnd_pos(tbl->record[0], rowid_a); + DBUG_ASSERT(!error); + error= tbl->file->ha_rnd_pos(tbl->record[1], rowid_b); + DBUG_ASSERT(!error); + /* + Compare the two rows by the corresponding values of the indexed + columns. + */ + for (uint i= 0; i < key_column_count; i++) + { + Field *cur_field= key_columns[i]->field; + if ((cmp_res= cur_field->cmp_offset(tbl->s->rec_buff_length))) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +int +Ordered_key::cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b) +{ + /* The result of comparing the two keys according to their row data. */ + int cmp_row_res= key->cmp_keys_by_row_data(*a, *b); + if (cmp_row_res) + return cmp_row_res; + return (*a < *b) ? -1 : (*a > *b) ? 1 : 0; +} + + +void Ordered_key::sort_keys() +{ + my_qsort2(key_buff, key_buff_elements, sizeof(rownum_t), + (qsort2_cmp) &cmp_keys_by_row_data_and_rownum, (void*) this); + /* Invalidate the current row position. */ + cur_key_idx= HA_POS_ERROR; +} + + +/* + The fraction of rows that do not contain NULL in the columns indexed by + this key. + + @retval 1 if there are no NULLs + @retval 0 if only NULLs +*/ + +double Ordered_key::null_selectivity() +{ + /* We should not be processing empty tables. */ + DBUG_ASSERT(tbl->file->stats.records); + return (1 - (double) null_count / (double) tbl->file->stats.records); +} + + +/* + Compare the value(s) of the current key in 'search_key' with the + data of the current table record. + + @notes The comparison result follows from the way compare_pred + is created in Ordered_key::init. Currently compare_pred compares + a field in of the current row with the corresponding Item that + contains the search key. + + @param row_num Number of the row (not index in the key_buff array) + + @retval -1 if (current row < search_key) + @retval 0 if (current row == search_key) + @retval +1 if (current row > search_key) +*/ + +int Ordered_key::cmp_key_with_search_key(rownum_t row_num) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tbl->file->ref_length; + uchar *cur_rowid= row_num_to_rowid + row_num * rowid_length; + int error, cmp_res; + + error= tbl->file->ha_rnd_pos(tbl->record[0], cur_rowid); + DBUG_ASSERT(!error); + + for (uint i= 0; i < key_column_count; i++) + { + cmp_res= compare_pred[i]->get_comparator()->compare(); + /* Unlike Arg_comparator::compare_row() here there should be no NULLs. */ + DBUG_ASSERT(!compare_pred[i]->null_value); + if (cmp_res) + return (cmp_res > 0 ? 1 : -1); + } + return 0; +} + + +/* + Find a key in a sorted array of keys via binary search. + + see create_subq_in_equalities() +*/ + +bool Ordered_key::lookup() +{ + DBUG_ASSERT(key_buff_elements); + + ha_rows lo= 0; + ha_rows hi= key_buff_elements - 1; + ha_rows mid; + int cmp_res; + + while (lo <= hi) + { + mid= lo + (hi - lo) / 2; + cmp_res= cmp_key_with_search_key(key_buff[mid]); + /* + In order to find the minimum match, check if the pevious element is + equal or smaller than the found one. If equal, we need to search further + to the left. + */ + if (!cmp_res && mid > 0) + cmp_res= !cmp_key_with_search_key(key_buff[mid - 1]) ? 1 : 0; + + if (cmp_res == -1) + { + /* row[mid] < search_key */ + lo= mid + 1; + } + else if (cmp_res == 1) + { + /* row[mid] > search_key */ + if (!mid) + goto not_found; + hi= mid - 1; + } + else + { + /* row[mid] == search_key */ + cur_key_idx= mid; + return TRUE; + } + } +not_found: + cur_key_idx= HA_POS_ERROR; + return FALSE; +} + + +/* + Move the current index pointer to the next key with the same column + values as the current key. Since the index is sorted, all such keys + are contiguous. +*/ + +bool Ordered_key::next_same() +{ + DBUG_ASSERT(key_buff_elements); + + if (cur_key_idx < key_buff_elements - 1) + { + /* + TIMOUR: + The below is quite inefficient, since as a result we will fetch every + row (except the last one) twice. There must be a more efficient way, + e.g. swapping record[0] and record[1], and reading only the new record. + */ + if (!cmp_keys_by_row_data(key_buff[cur_key_idx], key_buff[cur_key_idx + 1])) + { + ++cur_key_idx; + return TRUE; + } + } + return FALSE; +} + + +void Ordered_key::print(String *str) +{ + uint i; + str->append("{idx="); + str->qs_append(keyid); + str->append(", ("); + for (i= 0; i < key_column_count - 1; i++) + { + str->append(key_columns[i]->field->field_name); + str->append(", "); + } + str->append(key_columns[i]->field->field_name); + str->append("), "); + + str->append("null_bitmap: (bits="); + str->qs_append(null_key.n_bits); + str->append(", nulls= "); + str->qs_append((double)null_count); + str->append(", min_null= "); + str->qs_append((double)min_null_row); + str->append(", max_null= "); + str->qs_append((double)max_null_row); + str->append("), "); + + str->append('}'); +} + + +subselect_partial_match_engine::subselect_partial_match_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_engine(item_arg, result_arg), + tmp_table(tmp_table_arg), lookup_engine(engine_arg), + equi_join_conds(equi_join_conds_arg), + covering_null_row_width(covering_null_row_width_arg) +{} + + +int subselect_partial_match_engine::exec() +{ + Item_in_subselect *item_in= (Item_in_subselect *) item; + int res; + + /* Try to find a matching row by index lookup. */ + res= lookup_engine->copy_ref_key_simple(); + if (res == -1) + { + /* The result is FALSE based on the outer reference. */ + item_in->value= 0; + item_in->null_value= 0; + return 0; + } + else if (res == 0) + { + /* Search for a complete match. */ + if ((res= lookup_engine->index_lookup())) + { + /* An error occured during lookup(). */ + item_in->value= 0; + item_in->null_value= 0; + return res; + } + else if (item_in->value) + { + /* + A complete match was found, the result of IN is TRUE. + Notice: (this->item == lookup_engine->item) + */ + return 0; + } + } + + if (covering_null_row_width == tmp_table->s->fields) + { + /* + If there is a NULL-only row that coveres all columns the result of IN + is UNKNOWN. + */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + return 0; + } + + /* + There is no complete match. Look for a partial match (UNKNOWN result), or + no match (FALSE). + */ + if (tmp_table->file->inited) + tmp_table->file->ha_index_end(); + + if (partial_match()) + { + /* The result of IN is UNKNOWN. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 1; + item_in->null_value= 1; + } + else + { + /* The result of IN is FALSE. */ + item_in->value= 0; + /* + TIMOUR: which one is the right way to propagate an UNKNOWN result? + Should we also set empty_result_set= FALSE; ??? + */ + //item_in->was_null= 0; + item_in->null_value= 0; + } + + return 0; +} + + +void subselect_partial_match_engine::print(String *str, + enum_query_type query_type) +{ + /* + Should never be called as the actual engine cannot be known at query + optimization time. + */ + DBUG_ASSERT(FALSE); +} + + +/* + @param non_null_key_parts + @param partial_match_key_parts A union of all single-column NULL key parts. + @param count_partial_match_columns Number of NULL keyparts (set bits above). + + @retval FALSE the engine was initialized successfully + @retval TRUE there was some (memory allocation) error during initialization, + such errors should be interpreted as revert to other strategy +*/ + +bool +subselect_rowid_merge_engine::init(MY_BITMAP *non_null_key_parts, + MY_BITMAP *partial_match_key_parts) +{ + /* The length in bytes of the rowids (positions) of tmp_table. */ + uint rowid_length= tmp_table->file->ref_length; + ha_rows row_count= tmp_table->file->stats.records; + rownum_t cur_rownum= 0; + select_materialize_with_stats *result_sink= + (select_materialize_with_stats *) result; + uint cur_keyid= 0; + Item_in_subselect *item_in= (Item_in_subselect*) item; + int error; + + if (keys_count == 0) + { + /* There is nothing to initialize, we will only do regular lookups. */ + return FALSE; + } + + DBUG_ASSERT(!covering_null_row_width || (covering_null_row_width && + keys_count == 1 && + non_null_key_parts)); + /* + Allocate buffers to hold the merged keys and the mapping between rowids and + row numbers. + */ + if (!(merge_keys= (Ordered_key**) thd->alloc(keys_count * + sizeof(Ordered_key*))) || + !(row_num_to_rowid= (uchar*) my_malloc(row_count * rowid_length * + sizeof(uchar), MYF(MY_WME)))) + return TRUE; + + /* Create the only non-NULL key if there is any. */ + if (non_null_key_parts) + { + non_null_key= new Ordered_key(cur_keyid, tmp_table, item_in->left_expr, + 0, 0, 0, row_num_to_rowid); + if (non_null_key->init(non_null_key_parts)) + return TRUE; + merge_keys[cur_keyid]= non_null_key; + merge_keys[cur_keyid]->first(); + ++cur_keyid; + } + + /* + If there is a covering NULL row, the only key that is needed is the + only non-NULL key that is already created above. We create keys on + NULL-able columns only if there is no covering NULL row. + */ + if (!covering_null_row_width) + { + if (bitmap_init_memroot(&matching_keys, keys_count, thd->mem_root) || + bitmap_init_memroot(&matching_outer_cols, keys_count, thd->mem_root) || + bitmap_init_memroot(&null_only_columns, keys_count, thd->mem_root)) + return TRUE; + + /* + Create one single-column NULL-key for each column in + partial_match_key_parts. + */ + for (uint i= 0; i < partial_match_key_parts->n_bits; i++) + { + if (!bitmap_is_set(partial_match_key_parts, i)) + continue; + + if (result_sink->get_null_count_of_col(i) == row_count) + bitmap_set_bit(&null_only_columns, cur_keyid); + else + { + merge_keys[cur_keyid]= new Ordered_key( + cur_keyid, tmp_table, + item_in->left_expr->element_index(i), + result_sink->get_null_count_of_col(i), + result_sink->get_min_null_of_col(i), + result_sink->get_max_null_of_col(i), + row_num_to_rowid); + if (merge_keys[cur_keyid]->init(i)) + return TRUE; + merge_keys[cur_keyid]->first(); + } + ++cur_keyid; + } + } + + /* Populate the indexes with data from the temporary table. */ + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + tmp_table->null_row= 0; + while (TRUE) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error == HA_ERR_RECORD_DELETED) + { + /* We get this for duplicate records that should not be in tmp_table. */ + continue; + } + /* + This is a temp table that we fully own, there should be no other + cause to stop the iteration than EOF. + */ + DBUG_ASSERT(!error || error == HA_ERR_END_OF_FILE); + if (error == HA_ERR_END_OF_FILE) + { + DBUG_ASSERT(cur_rownum == tmp_table->file->stats.records); + break; + } + + /* + Save the position of this record in the row_num -> rowid mapping. + */ + tmp_table->file->position(tmp_table->record[0]); + memcpy(row_num_to_rowid + cur_rownum * rowid_length, + tmp_table->file->ref, rowid_length); + + /* Add the current row number to the corresponding keys. */ + if (non_null_key) + { + /* By definition there are no NULLs in the non-NULL key. */ + non_null_key->add_key(cur_rownum); + } + + for (uint i= (non_null_key ? 1 : 0); i < keys_count; i++) + { + /* + Check if the first and only indexed column contains NULL in the curent + row, and add the row number to the corresponding key. + */ + if (tmp_table->field[merge_keys[i]->get_field_idx(0)]->is_null()) + merge_keys[i]->set_null(cur_rownum); + else + merge_keys[i]->add_key(cur_rownum); + } + ++cur_rownum; + } + + tmp_table->file->ha_rnd_end(); + + /* Sort all the keys by their NULL selectivity. */ + my_qsort(merge_keys, keys_count, sizeof(Ordered_key*), + (qsort_cmp) cmp_keys_by_null_selectivity); + + /* Sort the keys in each of the indexes. */ + for (uint i= 0; i < keys_count; i++) + merge_keys[i]->sort_keys(); + + if (init_queue(&pq, keys_count, 0, FALSE, + subselect_rowid_merge_engine::cmp_keys_by_cur_rownum, NULL)) + return TRUE; + + return FALSE; +} + + +subselect_rowid_merge_engine::~subselect_rowid_merge_engine() +{ + /* None of the resources below is allocated if there are no ordered keys. */ + if (keys_count) + { + my_free((char*) row_num_to_rowid, MYF(0)); + for (uint i= 0; i < keys_count; i++) + delete merge_keys[i]; + delete_queue(&pq); + if (tmp_table->file->inited == handler::RND) + tmp_table->file->ha_rnd_end(); + } +} + + +void subselect_rowid_merge_engine::cleanup() +{ +} + + +/* + Quick sort comparison function to compare keys in order of decreasing bitmap + selectivity, so that the most selective keys come first. + + @param k1 first key to compare + @param k2 second key to compare + + @retval 1 if k1 is less selective than k2 + @retval 0 if k1 is equally selective as k2 + @retval -1 if k1 is more selective than k2 +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_null_selectivity(Ordered_key **k1, + Ordered_key **k2) +{ + double k1_sel= (*k1)->null_selectivity(); + double k2_sel= (*k2)->null_selectivity(); + if (k1_sel < k2_sel) + return 1; + if (k1_sel > k2_sel) + return -1; + return 0; +} + + +/* +*/ + +int +subselect_rowid_merge_engine::cmp_keys_by_cur_rownum(void *arg, + uchar *k1, uchar *k2) +{ + rownum_t r1= ((Ordered_key*) k1)->current(); + rownum_t r2= ((Ordered_key*) k2)->current(); + + return (r1 < r2) ? -1 : (r1 > r2) ? 1 : 0; +} + + +/* + Check if certain table row contains a NULL in all columns for which there is + no match in the corresponding value index. + + @retval TRUE if a NULL row exists + @retval FALSE otherwise +*/ + +bool subselect_rowid_merge_engine::test_null_row(rownum_t row_num) +{ + Ordered_key *cur_key; + uint cur_id; + for (uint i = 0; i < keys_count; i++) + { + cur_key= merge_keys[i]; + cur_id= cur_key->get_keyid(); + if (bitmap_is_set(&matching_keys, cur_id)) + { + /* + The key 'i' (with id 'cur_keyid') already matches a value in row 'row_num', + thus we skip it as it can't possibly match a NULL. + */ + continue; + } + if (!cur_key->is_null(row_num)) + return FALSE; + } + return TRUE; +} + + +/* + @retval TRUE there is a partial match (UNKNOWN) + @retval FALSE there is no match at all (FALSE) +*/ + +bool subselect_rowid_merge_engine::partial_match() +{ + Ordered_key *min_key; /* Key that contains the current minimum position. */ + rownum_t min_row_num; /* Current row number of min_key. */ + Ordered_key *cur_key; + rownum_t cur_row_num; + uint count_nulls_in_search_key= 0; + bool res= FALSE; + + /* If there is a non-NULL key, it must be the first key in the keys array. */ + DBUG_ASSERT(!non_null_key || (non_null_key && merge_keys[0] == non_null_key)); + + /* All data accesses during execution are via handler::ha_rnd_pos() */ + tmp_table->file->ha_rnd_init(0); + + /* Check if there is a match for the columns of the only non-NULL key. */ + if (non_null_key && !non_null_key->lookup()) + { + res= FALSE; + goto end; + } + + /* + If there is a NULL (sub)row that covers all NULL-able columns, + then there is a guranteed partial match, and we don't need to search + for the matching row. + */ + if (covering_null_row_width) + { + res= TRUE; + goto end; + } + + if (non_null_key) + queue_insert(&pq, (uchar *) non_null_key); + /* + Do not add the non_null_key, since it was already processed above. + */ + bitmap_clear_all(&matching_outer_cols); + for (uint i= test(non_null_key); i < keys_count; i++) + { + DBUG_ASSERT(merge_keys[i]->get_column_count() == 1); + if (merge_keys[i]->get_search_key(0)->is_null()) + { + ++count_nulls_in_search_key; + bitmap_set_bit(&matching_outer_cols, merge_keys[i]->get_keyid()); + } + else if (merge_keys[i]->lookup()) + queue_insert(&pq, (uchar *) merge_keys[i]); + } + + /* + If the outer reference consists of only NULLs, or if it has NULLs in all + nullable columns, the result is UNKNOWN. + */ + if (count_nulls_in_search_key == + ((Item_in_subselect *) item)->left_expr->cols() - + (non_null_key ? non_null_key->get_column_count() : 0)) + { + res= TRUE; + goto end; + } + + /* + If there is no NULL (sub)row that covers all NULL columns, and there is no + single match for any of the NULL columns, the result is FALSE. + */ + if (pq.elements - test(non_null_key) == 0) + { + res= FALSE; + goto end; + } + + DBUG_ASSERT(pq.elements); + + min_key= (Ordered_key*) queue_remove(&pq, 0); + min_row_num= min_key->current(); + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + if (min_key->next_same()) + queue_insert(&pq, (uchar *) min_key); + + if (pq.elements == 0) + { + /* + Check the only matching row of the only key min_key for NULL matches + in the other columns. + */ + res= test_null_row(min_row_num); + goto end; + } + + while (TRUE) + { + cur_key= (Ordered_key*) queue_remove(&pq, 0); + cur_row_num= cur_key->current(); + + if (cur_row_num == min_row_num) + bitmap_set_bit(&matching_keys, cur_key->get_keyid()); + else + { + /* Follows from the correct use of priority queue. */ + DBUG_ASSERT(cur_row_num > min_row_num); + if (test_null_row(min_row_num)) + { + res= TRUE; + goto end; + } + else + { + min_key= cur_key; + min_row_num= cur_row_num; + bitmap_copy(&matching_keys, &null_only_columns); + bitmap_set_bit(&matching_keys, min_key->get_keyid()); + bitmap_union(&matching_keys, &matching_outer_cols); + } + } + + if (cur_key->next_same()) + queue_insert(&pq, (uchar *) cur_key); + + if (pq.elements == 0) + { + /* Check the last row of the last column in PQ for NULL matches. */ + res= test_null_row(min_row_num); + goto end; + } + } + + /* We should never get here - all branches must be handled explicitly above. */ + DBUG_ASSERT(FALSE); + +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +subselect_table_scan_engine::subselect_table_scan_engine( + subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg) +{} + + +/* + TIMOUR: + This method is based on subselect_uniquesubquery_engine::scan_table(). + Consider refactoring somehow, 80% of the code is the same. + + for each row_i in tmp_table + { + count_matches= 0; + for each row element row_i[j] + { + if (outer_ref[j] is NULL || row_i[j] is NULL || outer_ref[j] == row_i[j]) + ++count_matches; + } + if (count_matches == outer_ref.elements) + return TRUE + } + return FALSE +*/ + +bool subselect_table_scan_engine::partial_match() +{ + List_iterator_fast<Item> equality_it(*equi_join_conds); + Item *cur_eq; + uint count_matches; + int error; + bool res; + + tmp_table->file->ha_rnd_init(1); + tmp_table->file->extra_opt(HA_EXTRA_CACHE, + current_thd->variables.read_buff_size); + /* + TIMOUR: + scan_table() also calls "table->null_row= 0;", why, do we need it? + */ + for (;;) + { + error= tmp_table->file->ha_rnd_next(tmp_table->record[0]); + if (error) { + if (error == HA_ERR_RECORD_DELETED) + { + error= 0; + continue; + } + if (error == HA_ERR_END_OF_FILE) + { + error= 0; + break; + } + else + { + error= report_error(tmp_table, error); + break; + } + } + + equality_it.rewind(); + count_matches= 0; + while ((cur_eq= equality_it++)) + { + DBUG_ASSERT(cur_eq->type() == Item::FUNC_ITEM && + ((Item_func*)cur_eq)->functype() == Item_func::EQ_FUNC); + if (!cur_eq->val_int() && !cur_eq->null_value) + break; + ++count_matches; + } + if (count_matches == tmp_table->s->fields) + { + res= TRUE; /* Found a matching row. */ + goto end; + } + } + + res= FALSE; +end: + tmp_table->file->ha_rnd_end(); + return res; +} + + +void subselect_table_scan_engine::cleanup() +{ +} === modified file 'sql/item_subselect.h' --- a/sql/item_subselect.h 2010-02-11 23:59:58 +0000 +++ b/sql/item_subselect.h 2010-03-09 10:14:06 +0000 @@ -297,7 +297,7 @@ Representation of IN subquery predicates of the form "left_expr IN (SELECT ...)". - @detail + @details This class has: - A "subquery execution engine" (as a subclass of Item_subselect) that allows it to evaluate subqueries. (and this class participates in execution by @@ -319,6 +319,12 @@ */ List<Cached_item> *left_expr_cache; bool first_execution; + /* + Set to TRUE if at query execution time we determine that this item's + value is a constant during this execution. We need this member because + it is not possible to substitute 'this' with a constant item. + */ + bool is_constant; /* expr & optimizer used in subselect rewriting to store Item for @@ -387,8 +393,8 @@ Item_in_subselect(Item * left_expr, st_select_lex *select_lex); Item_in_subselect() :Item_exists_subselect(), left_expr_cache(0), first_execution(TRUE), - optimizer(0), abort_on_null(0), pushed_cond_guards(NULL), - exec_method(NOT_TRANSFORMED), upper_item(0) + is_constant(FALSE), optimizer(0), abort_on_null(0), + pushed_cond_guards(NULL), exec_method(NOT_TRANSFORMED), upper_item(0) {} void cleanup(); subs_type substype() { return IN_SUBS; } @@ -421,6 +427,8 @@ void update_used_tables(); bool setup_engine(); bool init_left_expr_cache(); + /* Inform 'this' that it was computed, and contains a valid result. */ + void set_first_execution() { if (first_execution) first_execution= FALSE; } bool is_expensive_processor(uchar *arg); friend class Item_ref_null_helper; @@ -428,6 +436,7 @@ friend class Item_in_optimizer; friend class subselect_indexsubquery_engine; friend class subselect_hash_sj_engine; + friend class subselect_partial_match_engine; }; @@ -462,7 +471,8 @@ enum enum_engine_type {ABSTRACT_ENGINE, SINGLE_SELECT_ENGINE, UNION_ENGINE, UNIQUESUBQUERY_ENGINE, - INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE}; + INDEXSUBQUERY_ENGINE, HASH_SJ_ENGINE, + ROWID_MERGE_ENGINE, TABLE_SCAN_ENGINE}; subselect_engine(Item_subselect *si, select_result_interceptor *res) :thd(0) @@ -635,8 +645,10 @@ virtual void print (String *str, enum_query_type query_type); bool change_result(Item_subselect *si, select_result_interceptor *result); bool no_tables(); + int index_lookup(); /* TIMOUR: this method needs refactoring. */ int scan_table(); bool copy_ref_key(); + int copy_ref_key_simple(); /* TIMOUR: this method needs refactoring. */ bool no_rows() { return empty_result_set; } virtual enum_engine_type engine_type() { return UNIQUESUBQUERY_ENGINE; } }; @@ -705,50 +717,439 @@ /** - Compute an IN predicate via a hash semi-join. The subquery is materialized - during the first evaluation of the IN predicate. The IN predicate is executed - via the functionality inherited from subselect_uniquesubquery_engine. + Compute an IN predicate via a hash semi-join. This class is responsible for + the materialization of the subquery, and the selection of the correct and + optimal execution method (e.g. direct index lookup, or partial matching) for + the IN predicate. */ -class subselect_hash_sj_engine: public subselect_uniquesubquery_engine +class subselect_hash_sj_engine : public subselect_engine { protected: + /* The table into which the subquery is materialized. */ + TABLE *tmp_table; /* TRUE if the subquery was materialized into a temp table. */ bool is_materialized; /* The old engine already chosen at parse time and stored in permanent memory. Through this member we can re-create and re-prepare materialize_join for - each execution of a prepared statement. We akso resuse the functionality + each execution of a prepared statement. We also reuse the functionality of subselect_single_select_engine::[prepare | cols]. */ subselect_single_select_engine *materialize_engine; + /* The engine used to compute the IN predicate. */ + subselect_engine *lookup_engine; /* QEP to execute the subquery and materialize its result into a temporary table. Created during the first call to exec(). */ JOIN *materialize_join; - /* Temp table context of the outer select's JOIN. */ - TMP_TABLE_PARAM *tmp_param; + + /* Keyparts of the only non-NULL composite index in a rowid merge. */ + MY_BITMAP non_null_key_parts; + /* Keyparts of the single column indexes with NULL, one keypart per index. */ + MY_BITMAP partial_match_key_parts; + uint count_partial_match_columns; + uint count_null_only_columns; + /* + A conjunction of all the equality condtions between all pairs of expressions + that are arguments of an IN predicate. We need these to post-filter some + IN results because index lookups sometimes match values that are actually + not equal to the search key in SQL terms. + */ + Item_cond_and *semi_join_conds; + /* Possible execution strategies that can be used to compute hash semi-join.*/ + enum exec_strategy { + UNDEFINED, + COMPLETE_MATCH, /* Use regular index lookups. */ + PARTIAL_MATCH, /* Use some partial matching strategy. */ + PARTIAL_MATCH_MERGE, /* Use partial matching through index merging. */ + PARTIAL_MATCH_SCAN, /* Use partial matching through table scan. */ + IMPOSSIBLE /* Subquery materialization is not applicable. */ + }; + /* The chosen execution strategy. Computed after materialization. */ + exec_strategy strategy; +protected: + exec_strategy get_strategy_using_schema(); + exec_strategy get_strategy_using_data(); + size_t rowid_merge_buff_size(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + void choose_partial_match_strategy(bool has_non_null_key, + bool has_covering_null_row, + MY_BITMAP *partial_match_key_parts); + bool make_semi_join_conds(); + subselect_uniquesubquery_engine* make_unique_engine(); public: subselect_hash_sj_engine(THD *thd, Item_subselect *in_predicate, - subselect_single_select_engine *old_engine) - :subselect_uniquesubquery_engine(thd, NULL, in_predicate, NULL), - is_materialized(FALSE), materialize_engine(old_engine), - materialize_join(NULL), tmp_param(NULL) - {} + subselect_single_select_engine *old_engine) + :subselect_engine(in_predicate, NULL), tmp_table(NULL), + is_materialized(FALSE), materialize_engine(old_engine), lookup_engine(NULL), + materialize_join(NULL), count_partial_match_columns(0), + count_null_only_columns(0), semi_join_conds(NULL), strategy(UNDEFINED) + { + set_thd(thd); + } ~subselect_hash_sj_engine(); bool init_permanent(List<Item> *tmp_columns); bool init_runtime(); void cleanup(); - int prepare() { return 0; } + int prepare() { return 0; } /* Override virtual function in base class. */ int exec(); - virtual void print (String *str, enum_query_type query_type); + virtual void print(String *str, enum_query_type query_type); uint cols() { return materialize_engine->cols(); } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + table_map upper_select_const_tables() { return 0; } + bool no_rows() { return !tmp_table->file->stats.records; } virtual enum_engine_type engine_type() { return HASH_SJ_ENGINE; } -}; - + /* + TODO: factor out all these methods in a base subselect_index_engine class + because all of them have dummy implementations and should never be called. + */ + void fix_length_and_dec(Item_cache** row);//=>base class + void exclude(); //=>base class + //=>base class + bool change_result(Item_subselect *si, select_result_interceptor *result); + bool no_tables();//=>base class +}; + + +/* + Distinguish the type od (0-based) row numbers from the type of the index into + an array of row numbers. +*/ +typedef ha_rows rownum_t; + + +/* + An Ordered_key is an in-memory table index that allows O(log(N)) time + lookups of a multi-part key. + + If the index is over a single column, then this column may contain NULLs, and + the NULLs are stored and tested separately for NULL in O(1) via is_null(). + Multi-part indexes assume that the indexed columns do not contain NULLs. + + TODO: + = Due to the unnatural assymetry between single and multi-part indexes, it + makes sense to somehow refactor or extend the class. + + = This class can be refactored into a base abstract interface, and two + subclasses: + - one to represent single-column indexes, and + - another to represent multi-column indexes. + Such separation would allow slightly more efficient implementation of + the single-column indexes. + = The current design requires such indexes to be fully recreated for each + PS (re)execution, however most of the comprising objects can be reused. +*/ + +class Ordered_key : public Sql_alloc +{ +protected: + /* + Index of the key in an array of keys. This index allows to + construct (sub)sets of keys represented by bitmaps. + */ + uint keyid; + /* The table being indexed. */ + TABLE *tbl; + /* The columns being indexed. */ + Item_field **key_columns; + /* Number of elements in 'key_columns' (number of key parts). */ + uint key_column_count; + /* + An expression, or sequence of expressions that forms the search key. + The search key is a sequence when it is Item_row. Each element of the + sequence is accessible via Item::element_index(int i). + */ + Item *search_key; + +/* Value index related members. */ + /* + The actual value index, consists of a sorted sequence of row numbers. + */ + rownum_t *key_buff; + /* Number of elements in key_buff. */ + ha_rows key_buff_elements; + /* Current element in 'key_buff'. */ + ha_rows cur_key_idx; + /* + Mapping from row numbers to row ids. The element row_num_to_rowid[i] + contains a buffer with the rowid for the row numbered 'i'. + The memory for this member is not maintanined by this class because + all Ordered_key indexes of the same table share the same mapping. + */ + uchar *row_num_to_rowid; + /* + A sequence of predicates to compare the search key with the corresponding + columns of a table row from the index. + */ + Item_func_lt **compare_pred; + +/* Null index related members. */ + MY_BITMAP null_key; + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + +protected: + bool alloc_keys_buffers(); + /* + Quick sort comparison function that compares two rows of the same table + indentfied with their row numbers. + */ + int cmp_keys_by_row_data(rownum_t a, rownum_t b); + static int cmp_keys_by_row_data_and_rownum(Ordered_key *key, + rownum_t* a, rownum_t* b); + + int cmp_key_with_search_key(rownum_t row_num); + +public: + Ordered_key(uint keyid_arg, TABLE *tbl_arg, + Item *search_key_arg, ha_rows null_count_arg, + ha_rows min_null_row_arg, ha_rows max_null_row_arg, + uchar *row_num_to_rowid_arg); + ~Ordered_key(); + void cleanup(); + /* Initialize a multi-column index. */ + bool init(MY_BITMAP *columns_to_index); + /* Initialize a single-column index. */ + bool init(int col_idx); + + uint get_column_count() { return key_column_count; } + uint get_keyid() { return keyid; } + uint get_field_idx(uint i) + { + DBUG_ASSERT(i < key_column_count); + return key_columns[i]->field->field_index; + } + /* + Get the search key element that corresponds to the i-th key part of this + index. + */ + Item *get_search_key(uint i) + { + return search_key->element_index(key_columns[i]->field->field_index); + } + void add_key(rownum_t row_num) + { + /* The caller must know how many elements to add. */ + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + key_buff[cur_key_idx]= row_num; + ++cur_key_idx; + } + + void sort_keys(); + double null_selectivity(); + + /* + Position the current element at the first row that matches the key. + The key itself is propagated by evaluating the current value(s) of + this->search_key. + */ + bool lookup(); + /* Move the current index cursor to the first key. */ + void first() + { + DBUG_ASSERT(key_buff_elements); + cur_key_idx= 0; + } + /* TODO */ + bool next_same(); + /* Move the current index cursor to the next key. */ + bool next() + { + DBUG_ASSERT(key_buff_elements); + if (cur_key_idx < key_buff_elements - 1) + { + ++cur_key_idx; + return TRUE; + } + return FALSE; + }; + /* Return the current index element. */ + rownum_t current() + { + DBUG_ASSERT(key_buff_elements && cur_key_idx < key_buff_elements); + return key_buff[cur_key_idx]; + } + + void set_null(rownum_t row_num) + { + bitmap_set_bit(&null_key, row_num); + } + bool is_null(rownum_t row_num) + { + /* + Indexes consisting of only NULLs do not have a bitmap buffer at all. + Their only initialized member is 'n_bits', which is equal to the number + of temp table rows. + */ + if (null_count == tbl->file->stats.records) + { + DBUG_ASSERT(tbl->file->stats.records == null_key.n_bits); + return TRUE; + } + if (row_num > max_null_row || row_num < min_null_row) + return FALSE; + return bitmap_is_set(&null_key, row_num); + } + void print(String *str); +}; + + +class subselect_partial_match_engine : public subselect_engine +{ +protected: + /* The temporary table that contains a materialized subquery. */ + TABLE *tmp_table; + /* + The engine used to check whether an IN predicate is TRUE or not. If not + TRUE, then subselect_rowid_merge_engine further distinguishes between + FALSE and UNKNOWN. + */ + subselect_uniquesubquery_engine *lookup_engine; + /* A list of equalities between each pair of IN operands. */ + List<Item> *equi_join_conds; + /* + If there is a row, such that all its NULL-able components are NULL, this + member is set to the number of covered columns. If there is no covering + row, then this is 0. + */ + uint covering_null_row_width; +protected: + virtual bool partial_match()= 0; +public: + subselect_partial_match_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + int prepare() { return 0; } + int exec(); + void fix_length_and_dec(Item_cache**) {} + uint cols() { /* TODO: what is the correct value? */ return 1; } + uint8 uncacheable() { return UNCACHEABLE_DEPENDENT; } + void exclude() {} + table_map upper_select_const_tables() { return 0; } + bool change_result(Item_subselect*, select_result_interceptor*) + { DBUG_ASSERT(FALSE); return false; } + bool no_tables() { return false; } + bool no_rows() + { + /* + TODO: It is completely unclear what is the semantics of this + method. The current result is computed so that the call to no_rows() + from Item_in_optimizer::val_int() sets Item_in_optimizer::null_value + correctly. + */ + return !(((Item_in_subselect *) item)->null_value); + } + void print(String*, enum_query_type); + + friend void subselect_hash_sj_engine::cleanup(); +}; + + +class subselect_rowid_merge_engine: public subselect_partial_match_engine +{ +protected: + /* + Mapping from row numbers to row ids. The rowids are stored sequentially + in the array - rowid[i] is located in row_num_to_rowid + i * rowid_length. + */ + uchar *row_num_to_rowid; + /* + A subset of all the keys for which there is a match for the same row. + Used during execution. Computed for each outer reference + */ + MY_BITMAP matching_keys; + /* + The columns of the outer reference that are NULL. Computed for each + outer reference. + */ + MY_BITMAP matching_outer_cols; + /* + Columns that consist of only NULLs. Such columns match any value. + Computed once per query execution. + */ + MY_BITMAP null_only_columns; + /* + Indexes of row numbers, sorted by <column_value, row_number>. If an + index may contain NULLs, the NULLs are stored efficiently in a bitmap. + + The indexes are sorted by the selectivity of their NULL sub-indexes, the + one with the fewer NULLs is first. Thus, if there is any index on + non-NULL columns, it is contained in keys[0]. + */ + Ordered_key **merge_keys; + /* The number of elements in keys. */ + uint keys_count; + /* + An index on all non-NULL columns of 'tmp_table'. The index has the + logical form: <[v_i1 | ... | v_ik], rownum>. It allows to find the row + number where the columns c_i1,...,c1_k contain the values v_i1,...,v_ik. + If such an index exists, it is always the first element of 'keys'. + */ + Ordered_key *non_null_key; + /* + Priority queue of Ordered_key indexes, one per NULLable column. + This queue is used by the partial match algorithm in method exec(). + */ + QUEUE pq; +protected: + /* + Comparison function to compare keys in order of decreasing bitmap + selectivity. + */ + static int cmp_keys_by_null_selectivity(Ordered_key **k1, Ordered_key **k2); + /* + Comparison function used by the priority queue pq, the 'smaller' key + is the one with the smaller current row number. + */ + static int cmp_keys_by_cur_rownum(void *arg, uchar *k1, uchar *k2); + + bool test_null_row(rownum_t row_num); + bool partial_match(); +public: + subselect_rowid_merge_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, uint keys_count_arg, + uint covering_null_row_width_arg, + Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg) + :subselect_partial_match_engine(engine_arg, tmp_table_arg, item_arg, + result_arg, equi_join_conds_arg, + covering_null_row_width_arg), + keys_count(keys_count_arg), non_null_key(NULL) + { + thd= lookup_engine->get_thd(); + } + ~subselect_rowid_merge_engine(); + bool init(MY_BITMAP *non_null_key_parts, MY_BITMAP *partial_match_key_parts); + void cleanup(); + virtual enum_engine_type engine_type() { return ROWID_MERGE_ENGINE; } +}; + + +class subselect_table_scan_engine: public subselect_partial_match_engine +{ +protected: + bool partial_match(); +public: + subselect_table_scan_engine(subselect_uniquesubquery_engine *engine_arg, + TABLE *tmp_table_arg, Item_subselect *item_arg, + select_result_interceptor *result_arg, + List<Item> *equi_join_conds_arg, + uint covering_null_row_width_arg); + void cleanup(); + virtual enum_engine_type engine_type() { return TABLE_SCAN_ENGINE; } +}; === modified file 'sql/mysql_priv.h' --- a/sql/mysql_priv.h 2010-01-17 14:55:08 +0000 +++ b/sql/mysql_priv.h 2010-03-09 10:14:06 +0000 @@ -552,12 +552,14 @@ #define OPTIMIZER_SWITCH_LOOSE_SCAN 64 #define OPTIMIZER_SWITCH_MATERIALIZATION 128 #define OPTIMIZER_SWITCH_SEMIJOIN 256 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE 512 +#define OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN 1024 #ifdef DBUG_OFF -# define OPTIMIZER_SWITCH_LAST 512 +# define OPTIMIZER_SWITCH_LAST 2048 #else -# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 512 -# define OPTIMIZER_SWITCH_LAST 1024 +# define OPTIMIZER_SWITCH_TABLE_ELIMINATION 2048 +# define OPTIMIZER_SWITCH_LAST 4096 #endif #ifdef DBUG_OFF @@ -570,8 +572,10 @@ OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) -#else + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) +#else # define OPTIMIZER_SWITCH_DEFAULT (OPTIMIZER_SWITCH_INDEX_MERGE | \ OPTIMIZER_SWITCH_INDEX_MERGE_UNION | \ OPTIMIZER_SWITCH_INDEX_MERGE_SORT_UNION | \ @@ -581,7 +585,9 @@ OPTIMIZER_SWITCH_FIRSTMATCH | \ OPTIMIZER_SWITCH_LOOSE_SCAN | \ OPTIMIZER_SWITCH_MATERIALIZATION | \ - OPTIMIZER_SWITCH_SEMIJOIN) + OPTIMIZER_SWITCH_SEMIJOIN | \ + OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE|\ + OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN) #endif /* === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2010-01-17 14:55:08 +0000 +++ b/sql/mysqld.cc 2010-03-09 10:14:06 +0000 @@ -301,7 +301,9 @@ "index_merge","index_merge_union","index_merge_sort_union", "index_merge_intersection", "index_condition_pushdown", - "firstmatch","loosescan","materialization", "semijoin", + "firstmatch","loosescan","materialization", "semijoin", + "partial_match_rowid_merge", + "partial_match_table_scan", #ifndef DBUG_OFF "table_elimination", #endif @@ -320,6 +322,8 @@ sizeof("loosescan") - 1, sizeof("materialization") - 1, sizeof("semijoin") - 1, + sizeof("partial_match_rowid_merge") - 1, + sizeof("partial_match_table_scan") - 1, #ifndef DBUG_OFF sizeof("table_elimination") - 1, #endif @@ -5794,7 +5798,8 @@ OPT_RECORD_RND_BUFFER, OPT_DIV_PRECINCREMENT, OPT_RELAY_LOG_SPACE_LIMIT, OPT_RELAY_LOG_PURGE, OPT_SLAVE_NET_TIMEOUT, OPT_SLAVE_COMPRESSED_PROTOCOL, OPT_SLOW_LAUNCH_TIME, - OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_DEBUGGING, OPT_DEBUG_FLUSH, + OPT_SLAVE_TRANS_RETRIES, OPT_READONLY, OPT_ROWID_MERGE_BUFF_SIZE, + OPT_DEBUGGING, OPT_DEBUG_FLUSH, OPT_SORT_BUFFER, OPT_TABLE_OPEN_CACHE, OPT_TABLE_DEF_CACHE, OPT_THREAD_CONCURRENCY, OPT_THREAD_CACHE_SIZE, OPT_TMP_TABLE_SIZE, OPT_THREAD_STACK, @@ -7130,6 +7135,11 @@ (uchar**) &max_system_variables.range_alloc_block_size, 0, GET_ULONG, REQUIRED_ARG, RANGE_ALLOC_BLOCK_SIZE, RANGE_ALLOC_BLOCK_SIZE, (longlong) ULONG_MAX, 0, 1024, 0}, + {"rowid_merge_buff_size", OPT_ROWID_MERGE_BUFF_SIZE, + "The size of the buffers used [NOT] IN evaluation via partial matching.", + (uchar**) &global_system_variables.rowid_merge_buff_size, + (uchar**) &max_system_variables.rowid_merge_buff_size, 0, GET_ULONG, + REQUIRED_ARG, 8*1024*1024L, 0, MAX_MEM_TABLE_SIZE/2, 0, 1, 0}, {"read_buffer_size", OPT_RECORD_BUFFER, "Each thread that does a sequential scan allocates a buffer of this size for each table it scans. If you do many sequential scans, you may want to increase this value.", (uchar**) &global_system_variables.read_buff_size, === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-15 06:32:54 +0000 +++ b/sql/opt_subselect.cc 2010-03-15 15:09:35 +0000 @@ -187,10 +187,10 @@ does not call setup_subquery_materialization(). We could make SELECT ... FROM DUAL call that function but that doesn't seem to be the case that is worth handling. - 4. Subquery predicate is a top-level predicate - (this implies it is not negated) - TODO: this is a limitation that should be lifted once we - implement correct NULL semantics (WL#3830) + 4. Either the subquery predicate is a top-level predicate, or at + least one partial match strategy is enabled. If no partial match + strategy is enabled, then materialization cannot be used for + non-top-level queries because it cannot handle NULLs correctly. 5. Subquery is non-correlated TODO: This is an overly restrictive condition. It can be extended to: @@ -204,8 +204,8 @@ (*) The subquery must be part of a SELECT statement. The current condition also excludes multi-table update statements. - We have to determine whether we will perform subquery materialization - before calling the IN=>EXISTS transformation, so that we know whether to + Determine whether we will perform subquery materialization before + calling the IN=>EXISTS transformation, so that we know whether to perform the whole transformation or only that part of it which wraps Item_in_subselect in an Item_in_optimizer. */ @@ -215,12 +215,14 @@ select_lex->master_unit()->first_select()->leaf_tables && // 3 thd->lex->sql_command == SQLCOM_SELECT && // * select_lex->outer_select()->leaf_tables && // 3A - subquery_types_allow_materialization(in_subs)) + subquery_types_allow_materialization(in_subs) && + // psergey-todo: duplicated_subselect_card_check: where it's done? + (in_subs->is_top_level_item() || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) &&//4 + !in_subs->is_correlated && // 5 + in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 { - // psergey-todo: duplicated_subselect_card_check: where it's done? - if (in_subs->is_top_level_item() && // 4 - !in_subs->is_correlated && // 5 - in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 in_subs->exec_method= Item_in_subselect::MATERIALIZATION; } === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-12-22 12:49:15 +0000 +++ b/sql/set_var.cc 2010-03-09 10:14:06 +0000 @@ -540,6 +540,9 @@ static sys_var_thd_ulong sys_range_alloc_block_size(&vars, "range_alloc_block_size", &SV::range_alloc_block_size); +static sys_var_thd_ulong sys_rowid_merge_buff_size(&vars, "rowid_merge_buff_size", + &SV::rowid_merge_buff_size); + static sys_var_thd_ulong sys_query_alloc_block_size(&vars, "query_alloc_block_size", &SV::query_alloc_block_size, 0, fix_thd_mem_root); === modified file 'sql/sql_class.cc' --- a/sql/sql_class.cc 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.cc 2010-02-19 21:55:57 +0000 @@ -42,6 +42,7 @@ #include "sp_rcontext.h" #include "sp_cache.h" +#include "sql_select.h" /* declares create_tmp_table() */ /* The following is used to initialise Table_ident with a internal @@ -2877,6 +2878,71 @@ return 0; } + +bool +select_materialize_with_stats:: +create_result_table(THD *thd_arg, List<Item> *column_types, + bool is_union_distinct, ulonglong options, + const char *table_alias, bool bit_fields_as_long) +{ + DBUG_ASSERT(table == 0); + tmp_table_param.field_count= column_types->elements; + tmp_table_param.bit_fields_as_long= bit_fields_as_long; + + if (! (table= create_tmp_table(thd_arg, &tmp_table_param, *column_types, + (ORDER*) 0, is_union_distinct, 1, + options, HA_POS_ERROR, (char*) table_alias))) + return TRUE; + + col_stat= (Column_statistics*) table->in_use->alloc(table->s->fields * + sizeof(Column_statistics)); + if (!stat) + return TRUE; + + cleanup(); + + table->file->extra(HA_EXTRA_WRITE_CACHE); + table->file->extra(HA_EXTRA_IGNORE_DUP_KEY); + return FALSE; +} + + +/** + Override select_union::send_data to analyze each row for NULLs and to + update null_statistics before sending data to the client. + + @return TRUE if fatal error when sending data to the client + @return FALSE on success +*/ + +bool select_materialize_with_stats::send_data(List<Item> &items) +{ + List_iterator_fast<Item> item_it(items); + Item *cur_item; + Column_statistics *cur_col_stat= col_stat; + uint nulls_in_row= 0; + + ++count_rows; + + while ((cur_item= item_it++)) + { + if (cur_item->is_null()) + { + ++cur_col_stat->null_count; + cur_col_stat->max_null_row= count_rows; + if (!cur_col_stat->min_null_row) + cur_col_stat->min_null_row= count_rows; + ++nulls_in_row; + } + ++cur_col_stat; + } + if (nulls_in_row > max_nulls_in_row) + max_nulls_in_row= nulls_in_row; + + return select_union::send_data(items); +} + + /**************************************************************************** TMP_TABLE_PARAM ****************************************************************************/ === modified file 'sql/sql_class.h' --- a/sql/sql_class.h 2010-02-17 21:59:41 +0000 +++ b/sql/sql_class.h 2010-03-09 10:14:06 +0000 @@ -343,6 +343,8 @@ ulong mrr_buff_size; ulong div_precincrement; ulong sortbuff_size; + /* Total size of all buffers used by the subselect_rowid_merge_engine. */ + ulong rowid_merge_buff_size; ulong thread_handling; ulong tx_isolation; ulong completion_type; @@ -2740,19 +2742,20 @@ class select_union :public select_result_interceptor { +protected: TMP_TABLE_PARAM tmp_table_param; public: TABLE *table; - select_union() :table(0) {} + select_union() :table(0) { tmp_table_param.init(); } int prepare(List<Item> &list, SELECT_LEX_UNIT *u); bool send_data(List<Item> &items); bool send_eof(); bool flush(); - bool create_result_table(THD *thd, List<Item> *column_types, - bool is_distinct, ulonglong options, - const char *alias, bool bit_fields_as_long); + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); }; /* Base subselect interface class */ @@ -2776,6 +2779,74 @@ bool send_data(List<Item> &items); }; + +/* + This class specializes select_union to collect statistics about the + data stored in the temp table. Currently the class collects statistcs + about NULLs. +*/ + +class select_materialize_with_stats : public select_union +{ +protected: + class Column_statistics + { + public: + /* Count of NULLs per column. */ + ha_rows null_count; + /* The row number that contains the first NULL in a column. */ + ha_rows min_null_row; + /* The row number that contains the last NULL in a column. */ + ha_rows max_null_row; + }; + + /* Array of statistics data per column. */ + Column_statistics* col_stat; + + /* + The number of columns in the biggest sub-row that consists of only + NULL values. + */ + ha_rows max_nulls_in_row; + /* + Count of rows writtent to the temp table. This is redundant as it is + already stored in handler::stats.records, however that one is relatively + expensive to compute (given we need that for evry row). + */ + ha_rows count_rows; + +public: + select_materialize_with_stats() {} + virtual bool create_result_table(THD *thd, List<Item> *column_types, + bool is_distinct, ulonglong options, + const char *alias, bool bit_fields_as_long); + bool init_result_table(ulonglong select_options); + bool send_data(List<Item> &items); + void cleanup() + { + memset(col_stat, 0, table->s->fields * sizeof(Column_statistics)); + max_nulls_in_row= 0; + count_rows= 0; + } + ha_rows get_null_count_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].null_count; + } + ha_rows get_max_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].max_null_row; + } + ha_rows get_min_null_of_col(uint idx) + { + DBUG_ASSERT(idx < table->s->fields); + return col_stat[idx].min_null_row; + } + ha_rows get_max_nulls_in_row() { return max_nulls_in_row; } +}; + + /* used in independent ALL/ANY optimisation */ class select_max_min_finder_subselect :public select_subselect { === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-14 18:25:43 +0000 +++ b/sql/sql_select.cc 2010-03-15 14:34:56 +0000 @@ -874,6 +874,9 @@ { DBUG_PRINT("info",("No tables")); error= 0; + /* Create all structures needed for materialized subquery execution. */ + if (setup_subquery_materialization()) + DBUG_RETURN(1); DBUG_RETURN(0); } error= -1; // Error is sent to client @@ -11258,7 +11261,7 @@ param->group_buff=group_buff; share->keys=1; share->uniques= test(using_unique_constraint); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME; keyinfo->usable_key_parts=keyinfo->key_parts= param->group_parts; @@ -11344,7 +11347,7 @@ keyinfo->key_parts * sizeof(KEY_PART_INFO)))) goto err; bzero((void*) key_part_info, keyinfo->key_parts * sizeof(KEY_PART_INFO)); - table->key_info=keyinfo; + table->key_info= table->s->key_info= keyinfo; keyinfo->key_part=key_part_info; keyinfo->flags=HA_NOSAME | HA_NULL_ARE_EQUAL; keyinfo->key_length= 0; // Will compute the sum of the parts below.

1 0

Re: [Maria-developers] options for CREATE TABLE (MWL#43)
by Michael Widenius 15 Mar '10

15 Mar '10

Hi! >>>>> "Sergei" == Sergei Golubchik <serg(a)askmonty.org> writes: <cut> >>> 2. Unknown option should be an error by default. >> >> OK. The only problem is that it is contradict to Monty requirements. >> Our initial decision was issue error if option was added explicitly. >> The only problem is that it is very difficult to implement - we write >> options to .frm first then read them and pass to engine. I have no >> idea how to pass this information via/over frm. Sergei> I hope you've seen my reasoning below about optimizing for a common Sergei> case. Monty wants boundary cases to work - like changing engines back Sergei> and forth and replication. I am saying that by default unknown options Sergei> should be an error, but one should be able to disable that. Sergei> "An error if opion as added explicitly" does not solve all boundary Sergei> cases, for example, restoring a dump into a different engine. Sergei> Monty would probably want to cover that too. As almost all options are just 'extra information', I prefer that by default one doesn't get an error if the engine doesn't recognize the option. This is otherwise it's hell for automatic create table tools to work. It's much easier if one can just choose engine and then different options, some which are supported and others that may not be supported. Otherwise each tool would need to have a list of all existing engines and what options each support, which would be real hell. >>> 3. use something my_getopt-like as we discussed, don't force every >>> engine to parse its options >> >> I can add such function for users to use, but it will be thier choice >> use it or do not, is it OK? Sergei> What was the problem with doing it automatically ? Beccause the engine will still needs to do a switch over all options it supports, so it's hard to do it automatically. >>> 4. make options immutable to avoid copying them in ::clone >> >> I do not know way to do it if they should be allocated in different >> mem_roots. Sergei> Example ? Where are they allocated in different memroots ? This should work if we create the new table and reopen it before the old table is closed (which should be the case). >>> 5. don't check for changed options in alter table with your >>> check_if_incompatible_data. let the engine do that. >> >> This and 8 require big changes engine and ALTER TABLE. Monty's >> requirement was do not touch current code. I would be glad if you >> discuss it and make some non contradicting requirement. No comments, but I think this is easier to do on the top level than in the engine (but I don't remember Sanjas code exactly regarding this). >>> 7. parser: make the equal sign optional >> >> I have some doubts that it is doable >> >> DATA DIRECTORY TEST VALUE ... >> >> Does it mean: >> >> DATA = DIRECTORY TEST = VALUE ... >> >> or >> >> DATA DIRECTORY = TEST VALUE ... ? - error >> (ALTER TABLE uses create_table_options_space_separated list of options) Sergei> did you try the code from my previous email ? Agree with sanja that not having = can lead to parse problems. Also using = is more readable so I would prefer to over time start deprecate space between keyword and value. <cut> >>>> === modified file 'sql/sql_table.cc' >>>> --- sql/sql_table.cc 2010-02-12 08:47:31 +0000 >>>> +++ sql/sql_table.cc 2010-03-04 20:46:55 +0000 >>>> @@ -5789,6 +5791,15 @@ compare_tables(TABLE *table, >>>> DBUG_RETURN(0); >>>> } >>>> >>>> + if (!is_equal_create_options(tmp_new_field->create_options.first, >>>> + field->create_options.first)) >>>> + { >>> >>> I am not sure this should be checked on MySQL level, we don't know the >>> semantics of options. I'd say this check belong to >>> handler::check_if_incompatible_data() and should be implemented in the >>> storage engine internally. >> >> Monty even requested me to recreate .frm even if case of KEY was chenged >> (which is clear do not chengr semantic) - i.e. any change == rewriting >> .frm. So your requests contradict here it should be discussed (I do not see >> sens nor harm in such rewriting policy) Sergei> recreating frm is one thing, doing a full alter with copying the data is Sergei> another. I'm saying that it's not MySQL that should decide what change Sergei> in table options requires copy_data_between_tables - but the engine Sergei> itself. Agree that it's only the engine that knows if we need to copy the data or not. >>>> +plugin_option_value: >>>> + DEFAULT >>>> + { >>>> + $$.str= NULL; /* We are going to remove the option */ >>>> + $$.length= 0; >>>> + } >>>> + | NULL_SYM >>> >>> I don't like this trick. >>> If you don't support NULLs, dont't allow users to specify them >> >> how it can be stored as parameter value? Such semantic prevent users of >> thinking that assigning NULL will make it really NULL not "NULL". Sergei> It won't be "NULL", IDENT_sys that you use in plugin_option_value Sergei> will not treat NULL as an ident. I think if you simply remove Sergei> NULL alternative from the plugin_option_value rule, you'll end up Sergei> having a syntax error for option=NULL, which is better than what you Sergei> have now. Ok with me that we delete the =NULL syntax to remove options. >>>> +++ sql/sql_create_options.cc 2010-03-04 20:46:55 +0000 >>>> +my_bool create_option_add(CREATE_OPTION_LIST *options, MEM_ROOT *root, >>>> + const LEX_STRING *str_key, >>>> + const LEX_STRING *str_val, >>>> + my_bool *changed) >>>> +{ >>>> + CREATE_OPTION *cur_option, **option; >>>> + char *key, *val; >>>> + my_bool not_used; >>>> + my_bool copy= FALSE; >>>> + my_bool replace= FALSE; >>>> + DBUG_ENTER("create_option_add"); >>>> + DBUG_PRINT("enter", ("key: '%s' value: '%s'", >>>> + str_key->str, str_val->str)); >>>> + if (changed) >>>> + copy= TRUE; >>>> + else >>>> + changed= &not_used; >>>> + >>>> + DBUG_ASSERT(options->first || >>>> + (!options->first && options->last == &options->first)); >>>> + *changed= FALSE; >>> >>> Hmm, strange. From the way you use 'changed' I thought it should >>> accumulate >>> the results - I mean, it's one variable that is passed into >>> create_option_add() for all options. Apparently at the end it should be >>> true if *any* of the options has changed. >>> >>> But then, why do you set it to false inside create_option_add() ? >> >> It was special case for call from ALTER TABLE and from parser. Only ALTER >> TABLE was interested in changes and so required copying parameters. Sergei> I don't understand. I also in my review thought it would be much more logical if 'changed' would be reset (if needed) on the outer level, not in the function. >>>> + >>>> + /* try to find the option first */ >>>> + for (option= &(options->first); >>>> + *option && my_strcasecmp(system_charset_info, >>>> + str_key->str, (*option)->key.str); >>>> + option= &((*option)->next)) ; >>>> + if (str_val->str) >>>> + { >>>> + /* add / replace */ >>>> + if (*option) >>>> + { >>>> + /* replace */ >>>> + cur_option= *option; >>>> + if (!(*changed) && >>>> + (cur_option->val.length != str_val->length || >>>> + memcmp(cur_option->val.str, str_val->str, str_val->length))) >>>> + { >>>> + *changed= TRUE; >>>> + } >>>> + replace= TRUE; >>>> + } >>>> + else >>>> + { Sergei> ... >>>> +CREATE_OPTION_LIST *create_create_options_array(MEM_ROOT *root, uint n) >>> >>> "create_create" is not a good name :( >> >> I did not found better but open for suggestion. Sergei> make_create_options_array ? Sergei> construct_create_options_array ? construct_create_options_array sounds nice to me. >>>> +my_bool create_options_read(const uchar *buff, uint length, MEM_ROOT >>>> *root, >>>> + TABLE_OPTIONS *opt) >>>> +{ >>>> + const uchar *buff_end= buff + length; >>>> + DBUG_ENTER("create_options_read"); >>>> + while (buff < buff_end) >>>> + { >>>> + CREATE_OPTION *option; >>>> + CREATE_OPTION_TYPES type; >>>> + uint index= 0; >>>> + >>>> + if (!(option= (CREATE_OPTION *) alloc_root(root, >>>> sizeof(CREATE_OPTION)))) >>>> + DBUG_RETURN(TRUE); >>>> + >>>> + DBUG_ASSERT(buff + 4 <= buff_end); >>>> + option->val.length= uint2korr(buff); >>>> + option->key.length= buff[2]; >>>> + option->next= NULL; >>>> + type= (CREATE_OPTION_TYPES)buff[3]; >>>> + buff+= 4; >>>> + switch (type) { >>>> + case CREATE_OPTION_FIELD: >>> >>> interesting encoding. so basically you support the case when field, >>> key, and table options are all written interleaved: >>> >>> <table option><key 1 option><field 5 option><table option><field 3 option> <key 4 option>... >>> >>> why the heck do you want to support it ? >> >> Could you propose other encoding taking into account that some fields, keys >> and tables do not have parameters and some has several ones? Sergei> Sure. Many :) Sergei> For example Sergei> <number of table options> Sergei> <length-encoded strings for table options> Sergei> <number of field 1 options> Sergei> <length-encoded strings for field 1 options> Sergei> <number of field 2 options> Sergei> <length-encoded strings for field 2 options> Sergei> ... Sergei> <number of key 1 options> Sergei> <length-encoded strings for key 1 options> Sergei> <number of key 2 options> Sergei> <length-encoded strings for key 2 options> Sergei> Assuming a table with three fields and two keys that would be Sergei> 0x02 0x05 "topt1" 0x03 "val" 0x03 "to2" 0x04 "val2" Sergei> 0x00 Sergei> 0x01 0x04 "fil1" 0x01 "1" Sergei> 0x03 0x01 "A" 0x02 "bb" 0x01 "B" 0x02 "CC" 0x02 "de" 0x01 "0" Sergei> 0x01 0x06 "packed" 0x03 "yes" Sergei> 0x00 I also originally thought about this (I would probably have stored things the above way if I would have coded this). However, I am not sure that the code would be shorter than Sanjas code. The fact that the code can handle cases that never happens in reality didn't bother me. Regards, Monty

1 0

[Maria-developers] Progress (by Knielsen): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 25 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Knielsen - Mon, 15 Mar 2010, 14:28)=-=- Research into the problem, and discussions on phone/mailing list Worked 25 hours and estimate 0 hours remain (original estimate increased by 25 hours). -=-=(Guest - Mon, 15 Mar 2010, 14:18)=-=- High-Level Specification modified. --- /tmp/wklog.107.old.9086 2010-03-15 14:18:18.000000000 +0000 +++ /tmp/wklog.107.new.9086 2010-03-15 14:18:18.000000000 +0000 @@ -1 +1,43 @@ +Current ideas/status after discussions on the mailing list: + + - Implement a set of plugin APIs and use them to move all of the existing + MySQL replication into a (set of) plugins. + + - Design the APIs so that they can support full MySQL replication, but also + so that they do not hardcode assumptions about how this replication + implementation is done, and so that they will be suitable for other types of + replication (Tungsten, Galera, parallel replication, ...). + + - APIs need to include the concept of a global transaction ID. Need to + determine the extent to which the semantics of such ID will be defined + by the API, and to which extend it will be defined by the plugin + implementations. + + - APIs should properly support reliable crash-recovery with decent + performance (eg. not require multiple mandatory fsync()s per commit, and + not make group commit impossible). + + - Would be nice if the API provided facilities for implementing good + consistency checking support (mainly checking master tables against slave + tables is hard here I think, but also applying wrong binlog data and + individual event checksums). + + +Steps to make this more concrete: + + - Investigate the current MySQL replication, and list all of the places where + a plugin implementation will need to connect/hook into the MySQL server. + * handler::{write,update,delete}_row() + * Statement execution + * Transaction start/commit + * Table open + * Query safe/not/safe for statement based replication + * Statement-based logging details (user variables, random seed, etc.) + * ... + + - Use this list to make an initial sketch of the set of APIs we need. + + - Use the list to determine the feasibility of this project and the level of + detail in the API needed to support a full replication implementation as a + plugin. -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html HIGH-LEVEL SPECIFICATION: Current ideas/status after discussions on the mailing list: - Implement a set of plugin APIs and use them to move all of the existing MySQL replication into a (set of) plugins. - Design the APIs so that they can support full MySQL replication, but also so that they do not hardcode assumptions about how this replication implementation is done, and so that they will be suitable for other types of replication (Tungsten, Galera, parallel replication, ...). - APIs need to include the concept of a global transaction ID. Need to determine the extent to which the semantics of such ID will be defined by the API, and to which extend it will be defined by the plugin implementations. - APIs should properly support reliable crash-recovery with decent performance (eg. not require multiple mandatory fsync()s per commit, and not make group commit impossible). - Would be nice if the API provided facilities for implementing good consistency checking support (mainly checking master tables against slave tables is hard here I think, but also applying wrong binlog data and individual event checksums). Steps to make this more concrete: - Investigate the current MySQL replication, and list all of the places where a plugin implementation will need to connect/hook into the MySQL server. * handler::{write,update,delete}_row() * Statement execution * Transaction start/commit * Table open * Query safe/not/safe for statement based replication * Statement-based logging details (user variables, random seed, etc.) * ... - Use this list to make an initial sketch of the set of APIs we need. - Use the list to determine the feasibility of this project and the level of detail in the API needed to support a full replication implementation as a plugin. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Progress (by Knielsen): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 25 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Knielsen - Mon, 15 Mar 2010, 14:28)=-=- Research into the problem, and discussions on phone/mailing list Worked 25 hours and estimate 0 hours remain (original estimate increased by 25 hours). -=-=(Guest - Mon, 15 Mar 2010, 14:18)=-=- High-Level Specification modified. --- /tmp/wklog.107.old.9086 2010-03-15 14:18:18.000000000 +0000 +++ /tmp/wklog.107.new.9086 2010-03-15 14:18:18.000000000 +0000 @@ -1 +1,43 @@ +Current ideas/status after discussions on the mailing list: + + - Implement a set of plugin APIs and use them to move all of the existing + MySQL replication into a (set of) plugins. + + - Design the APIs so that they can support full MySQL replication, but also + so that they do not hardcode assumptions about how this replication + implementation is done, and so that they will be suitable for other types of + replication (Tungsten, Galera, parallel replication, ...). + + - APIs need to include the concept of a global transaction ID. Need to + determine the extent to which the semantics of such ID will be defined + by the API, and to which extend it will be defined by the plugin + implementations. + + - APIs should properly support reliable crash-recovery with decent + performance (eg. not require multiple mandatory fsync()s per commit, and + not make group commit impossible). + + - Would be nice if the API provided facilities for implementing good + consistency checking support (mainly checking master tables against slave + tables is hard here I think, but also applying wrong binlog data and + individual event checksums). + + +Steps to make this more concrete: + + - Investigate the current MySQL replication, and list all of the places where + a plugin implementation will need to connect/hook into the MySQL server. + * handler::{write,update,delete}_row() + * Statement execution + * Transaction start/commit + * Table open + * Query safe/not/safe for statement based replication + * Statement-based logging details (user variables, random seed, etc.) + * ... + + - Use this list to make an initial sketch of the set of APIs we need. + + - Use the list to determine the feasibility of this project and the level of + detail in the API needed to support a full replication implementation as a + plugin. -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html HIGH-LEVEL SPECIFICATION: Current ideas/status after discussions on the mailing list: - Implement a set of plugin APIs and use them to move all of the existing MySQL replication into a (set of) plugins. - Design the APIs so that they can support full MySQL replication, but also so that they do not hardcode assumptions about how this replication implementation is done, and so that they will be suitable for other types of replication (Tungsten, Galera, parallel replication, ...). - APIs need to include the concept of a global transaction ID. Need to determine the extent to which the semantics of such ID will be defined by the API, and to which extend it will be defined by the plugin implementations. - APIs should properly support reliable crash-recovery with decent performance (eg. not require multiple mandatory fsync()s per commit, and not make group commit impossible). - Would be nice if the API provided facilities for implementing good consistency checking support (mainly checking master tables against slave tables is hard here I think, but also applying wrong binlog data and individual event checksums). Steps to make this more concrete: - Investigate the current MySQL replication, and list all of the places where a plugin implementation will need to connect/hook into the MySQL server. * handler::{write,update,delete}_row() * Statement execution * Transaction start/commit * Table open * Query safe/not/safe for statement based replication * Statement-based logging details (user variables, random seed, etc.) * ... - Use this list to make an initial sketch of the set of APIs we need. - Use the list to determine the feasibility of this project and the level of detail in the API needed to support a full replication implementation as a plugin. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Guest - Mon, 15 Mar 2010, 14:18)=-=- High-Level Specification modified. --- /tmp/wklog.107.old.9086 2010-03-15 14:18:18.000000000 +0000 +++ /tmp/wklog.107.new.9086 2010-03-15 14:18:18.000000000 +0000 @@ -1 +1,43 @@ +Current ideas/status after discussions on the mailing list: + + - Implement a set of plugin APIs and use them to move all of the existing + MySQL replication into a (set of) plugins. + + - Design the APIs so that they can support full MySQL replication, but also + so that they do not hardcode assumptions about how this replication + implementation is done, and so that they will be suitable for other types of + replication (Tungsten, Galera, parallel replication, ...). + + - APIs need to include the concept of a global transaction ID. Need to + determine the extent to which the semantics of such ID will be defined + by the API, and to which extend it will be defined by the plugin + implementations. + + - APIs should properly support reliable crash-recovery with decent + performance (eg. not require multiple mandatory fsync()s per commit, and + not make group commit impossible). + + - Would be nice if the API provided facilities for implementing good + consistency checking support (mainly checking master tables against slave + tables is hard here I think, but also applying wrong binlog data and + individual event checksums). + + +Steps to make this more concrete: + + - Investigate the current MySQL replication, and list all of the places where + a plugin implementation will need to connect/hook into the MySQL server. + * handler::{write,update,delete}_row() + * Statement execution + * Transaction start/commit + * Table open + * Query safe/not/safe for statement based replication + * Statement-based logging details (user variables, random seed, etc.) + * ... + + - Use this list to make an initial sketch of the set of APIs we need. + + - Use the list to determine the feasibility of this project and the level of + detail in the API needed to support a full replication implementation as a + plugin. -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html HIGH-LEVEL SPECIFICATION: Current ideas/status after discussions on the mailing list: - Implement a set of plugin APIs and use them to move all of the existing MySQL replication into a (set of) plugins. - Design the APIs so that they can support full MySQL replication, but also so that they do not hardcode assumptions about how this replication implementation is done, and so that they will be suitable for other types of replication (Tungsten, Galera, parallel replication, ...). - APIs need to include the concept of a global transaction ID. Need to determine the extent to which the semantics of such ID will be defined by the API, and to which extend it will be defined by the plugin implementations. - APIs should properly support reliable crash-recovery with decent performance (eg. not require multiple mandatory fsync()s per commit, and not make group commit impossible). - Would be nice if the API provided facilities for implementing good consistency checking support (mainly checking master tables against slave tables is hard here I think, but also applying wrong binlog data and individual event checksums). Steps to make this more concrete: - Investigate the current MySQL replication, and list all of the places where a plugin implementation will need to connect/hook into the MySQL server. * handler::{write,update,delete}_row() * Statement execution * Transaction start/commit * Table open * Query safe/not/safe for statement based replication * Statement-based logging details (user variables, random seed, etc.) * ... - Use this list to make an initial sketch of the set of APIs we need. - Use the list to determine the feasibility of this project and the level of detail in the API needed to support a full replication implementation as a plugin. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Guest - Mon, 15 Mar 2010, 14:18)=-=- High-Level Specification modified. --- /tmp/wklog.107.old.9086 2010-03-15 14:18:18.000000000 +0000 +++ /tmp/wklog.107.new.9086 2010-03-15 14:18:18.000000000 +0000 @@ -1 +1,43 @@ +Current ideas/status after discussions on the mailing list: + + - Implement a set of plugin APIs and use them to move all of the existing + MySQL replication into a (set of) plugins. + + - Design the APIs so that they can support full MySQL replication, but also + so that they do not hardcode assumptions about how this replication + implementation is done, and so that they will be suitable for other types of + replication (Tungsten, Galera, parallel replication, ...). + + - APIs need to include the concept of a global transaction ID. Need to + determine the extent to which the semantics of such ID will be defined + by the API, and to which extend it will be defined by the plugin + implementations. + + - APIs should properly support reliable crash-recovery with decent + performance (eg. not require multiple mandatory fsync()s per commit, and + not make group commit impossible). + + - Would be nice if the API provided facilities for implementing good + consistency checking support (mainly checking master tables against slave + tables is hard here I think, but also applying wrong binlog data and + individual event checksums). + + +Steps to make this more concrete: + + - Investigate the current MySQL replication, and list all of the places where + a plugin implementation will need to connect/hook into the MySQL server. + * handler::{write,update,delete}_row() + * Statement execution + * Transaction start/commit + * Table open + * Query safe/not/safe for statement based replication + * Statement-based logging details (user variables, random seed, etc.) + * ... + + - Use this list to make an initial sketch of the set of APIs we need. + + - Use the list to determine the feasibility of this project and the level of + detail in the API needed to support a full replication implementation as a + plugin. -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html HIGH-LEVEL SPECIFICATION: Current ideas/status after discussions on the mailing list: - Implement a set of plugin APIs and use them to move all of the existing MySQL replication into a (set of) plugins. - Design the APIs so that they can support full MySQL replication, but also so that they do not hardcode assumptions about how this replication implementation is done, and so that they will be suitable for other types of replication (Tungsten, Galera, parallel replication, ...). - APIs need to include the concept of a global transaction ID. Need to determine the extent to which the semantics of such ID will be defined by the API, and to which extend it will be defined by the plugin implementations. - APIs should properly support reliable crash-recovery with decent performance (eg. not require multiple mandatory fsync()s per commit, and not make group commit impossible). - Would be nice if the API provided facilities for implementing good consistency checking support (mainly checking master tables against slave tables is hard here I think, but also applying wrong binlog data and individual event checksums). Steps to make this more concrete: - Investigate the current MySQL replication, and list all of the places where a plugin implementation will need to connect/hook into the MySQL server. * handler::{write,update,delete}_row() * Statement execution * Transaction start/commit * Table open * Query safe/not/safe for statement based replication * Statement-based logging details (user variables, random seed, etc.) * ... - Use this list to make an initial sketch of the set of APIs we need. - Use the list to determine the feasibility of this project and the level of detail in the API needed to support a full replication implementation as a plugin. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Serg): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Serg): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: Sergei CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Serg - Mon, 15 Mar 2010, 14:13)=-=- Observers changed: Sergei DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Knielsen): New replication APIs (107)
by worklog-noreply＠askmonty.org 15 Mar '10

15 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: New replication APIs CREATION DATE..: Mon, 15 Mar 2010, 13:55 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 107 (http://askmonty.org/worklog/?tid=107) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: This is a top-level task for the project of designing a new set of replication APIs for MariaDB. This task is for the initial discussion of what to do and where to focus. The project is started in this email thread: https://lists.launchpad.net/maria-developers/msg01998.html ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Rev 2778: Merge in file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/
by Sergey Petrunya 15 Mar '10

15 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/ ------------------------------------------------------------ revno: 2778 [merge] revision-id: psergey(a)askmonty.org-20100315063535-jsp4jgya6lfqt8e6 parent: psergey(a)askmonty.org-20100315063254-z1ctm7srl0573s5c parent: psergey(a)askmonty.org-20100315060659-0spqc4jdav12ja2u committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7-rel timestamp: Mon 2010-03-15 09:35:35 +0300 message: Merge modified: mysql-test/r/type_datetime.result sp1f-type_datetime.result-20001228015634-jrgwqpilnfn4kvdp6wm5hp5imvf3tkek === modified file 'mysql-test/r/type_datetime.result' --- a/mysql-test/r/type_datetime.result 2010-02-11 21:59:32 +0000 +++ b/mysql-test/r/type_datetime.result 2010-03-15 06:06:59 +0000 @@ -516,7 +516,7 @@ 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables Warnings: Note 1276 Field or reference 'test.t1.cur_date' of SELECT #2 was resolved in SELECT #1 -Note 1003 select '1' AS `id`,'2007-04-25 18:30:22' AS `cur_date` from `test`.`t1` `x1` join `test`.`t1` where (('2007-04-25 18:30:22' = 0)) +Note 1003 select '1' AS `id`,'2007-04-25 18:30:22' AS `cur_date` from `test`.`t1` semi join (`test`.`t1` `x1`) where (('2007-04-25 18:30:22' = 0)) select * from t1 where id in (select id from t1 as x1 where (t1.cur_date is null)); id cur_date @@ -527,7 +527,7 @@ 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables Warnings: Note 1276 Field or reference 'test.t2.cur_date' of SELECT #2 was resolved in SELECT #1 -Note 1003 select '1' AS `id`,'2007-04-25' AS `cur_date` from `test`.`t2` `x1` join `test`.`t2` where (('2007-04-25' = 0)) +Note 1003 select '1' AS `id`,'2007-04-25' AS `cur_date` from `test`.`t2` semi join (`test`.`t2` `x1`) where (('2007-04-25' = 0)) select * from t2 where id in (select id from t2 as x1 where (t2.cur_date is null)); id cur_date

1 0

[Maria-developers] Rev 2777: Apply fix by Roy Lyseng: in file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/
by Sergey Petrunya 15 Mar '10

15 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/ ------------------------------------------------------------ revno: 2777 revision-id: psergey(a)askmonty.org-20100315063254-z1ctm7srl0573s5c parent: psergey(a)askmonty.org-20100314182543-4t3ehit7df20adu8 committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7-rel timestamp: Mon 2010-03-15 09:32:54 +0300 message: Apply fix by Roy Lyseng: Bug#48623: Multiple subqueries are optimized incorrectly The function setup_semijoin_dups_elimination() has a major loop that goes through every table in the JOIN object. Usually, there is a normal "plus one" increment in the for loop that implements this, but each semijoin nest is treated as one entity and there is another increment that skips past the semijoin nest to the next table in the JOIN object. However, when combining these two increments, the next joined table is skipped, and if that happens to be the start of another semijoin nest, the correct processing for that nest will not be carried out. === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-14 18:25:43 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-15 06:32:54 +0000 @@ -1079,3 +1079,36 @@ partner_id partner2 drop table t1,t2,t3,t4; +# +# Bug#48623 Multiple subqueries are optimized incorrectly +# +CREATE TABLE t1(val VARCHAR(10)); +CREATE TABLE t2(val VARCHAR(10)); +CREATE TABLE t3(val VARCHAR(10)); +INSERT INTO t1 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t2 VALUES('aaa'), ('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t3 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +EXPLAIN +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 +WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') +AND t1.val IN (SELECT t3.val FROM t3 +WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); +id select_type table type possible_keys key key_len ref rows Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 5 +1 PRIMARY t3 ALL NULL NULL NULL NULL 5 Using where; FirstMatch(t1) +1 PRIMARY t2 ALL NULL NULL NULL NULL 6 Using where; FirstMatch(t3) +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 +WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') +AND t1.val IN (SELECT t3.val FROM t3 +WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); +val +aaa +eee +DROP TABLE t1; +DROP TABLE t2; +DROP TABLE t3; +# End of Bug#48623 === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-14 18:25:43 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-15 06:32:54 +0000 @@ -1083,6 +1083,39 @@ partner_id partner2 drop table t1,t2,t3,t4; +# +# Bug#48623 Multiple subqueries are optimized incorrectly +# +CREATE TABLE t1(val VARCHAR(10)); +CREATE TABLE t2(val VARCHAR(10)); +CREATE TABLE t3(val VARCHAR(10)); +INSERT INTO t1 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t2 VALUES('aaa'), ('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t3 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +EXPLAIN +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 +WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') +AND t1.val IN (SELECT t3.val FROM t3 +WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); +id select_type table type possible_keys key key_len ref rows Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 5 +1 PRIMARY t3 ALL NULL NULL NULL NULL 5 Using where; FirstMatch(t1); Using join buffer +1 PRIMARY t2 ALL NULL NULL NULL NULL 6 Using where; FirstMatch(t3); Using join buffer +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 +WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') +AND t1.val IN (SELECT t3.val FROM t3 +WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); +val +aaa +eee +DROP TABLE t1; +DROP TABLE t2; +DROP TABLE t3; +# End of Bug#48623 # # BUG#49129: Wrong result with IN-subquery with join_cache_level=6 and firstmatch=off # === modified file 'mysql-test/t/subselect_sj.test' --- a/mysql-test/t/subselect_sj.test 2010-03-14 18:25:43 +0000 +++ b/mysql-test/t/subselect_sj.test 2010-03-15 06:32:54 +0000 @@ -943,5 +943,35 @@ execute stmt; drop table t1,t2,t3,t4; - - +--echo # +--echo # Bug#48623 Multiple subqueries are optimized incorrectly +--echo # + +CREATE TABLE t1(val VARCHAR(10)); +CREATE TABLE t2(val VARCHAR(10)); +CREATE TABLE t3(val VARCHAR(10)); + +INSERT INTO t1 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t2 VALUES('aaa'), ('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); +INSERT INTO t3 VALUES('aaa'), ('bbb'), ('eee'), ('mmm'), ('ppp'); + +EXPLAIN +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 + WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') + AND t1.val IN (SELECT t3.val FROM t3 + WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); + +SELECT * +FROM t1 +WHERE t1.val IN (SELECT t2.val FROM t2 + WHERE t2.val LIKE 'a%' OR t2.val LIKE 'e%') + AND t1.val IN (SELECT t3.val FROM t3 + WHERE t3.val LIKE 'a%' OR t3.val LIKE 'e%'); + +DROP TABLE t1; +DROP TABLE t2; +DROP TABLE t3; + +--echo # End of Bug#48623 === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-14 18:25:43 +0000 +++ b/sql/opt_subselect.cc 2010-03-15 06:32:54 +0000 @@ -3030,7 +3030,7 @@ THD *thd= join->thd; DBUG_ENTER("setup_semijoin_dups_elimination"); - for (i= join->const_tables ; i < join->tables ; i++) + for (i= join->const_tables ; i < join->tables; ) { JOIN_TAB *tab=join->join_tab + i; POSITION *pos= join->best_positions + i; @@ -3039,7 +3039,7 @@ case SJ_OPT_MATERIALIZE: case SJ_OPT_MATERIALIZE_SCAN: /* Do nothing */ - i += pos->n_sj_tables; + i+= pos->n_sj_tables; break; case SJ_OPT_LOOSE_SCAN: { @@ -3055,7 +3055,7 @@ tab->loosescan_key_len= keylen; if (pos->n_sj_tables > 1) tab[pos->n_sj_tables - 1].do_firstmatch= tab; - i += pos->n_sj_tables; + i+= pos->n_sj_tables; break; } case SJ_OPT_DUPS_WEEDOUT: @@ -3152,7 +3152,7 @@ join->join_tab[first_table].flush_weedout_table= sjtbl; join->join_tab[i + pos->n_sj_tables - 1].check_weed_out_table= sjtbl; - i += pos->n_sj_tables; + i+= pos->n_sj_tables; break; } case SJ_OPT_FIRST_MATCH: @@ -3174,10 +3174,11 @@ } } j[-1].do_firstmatch= jump_to; - i += pos->n_sj_tables; + i+= pos->n_sj_tables; break; } case SJ_OPT_NONE: + i++; break; } }

1 0

[Maria-developers] Rev 2777: Update test results for the previous push in file:///home/psergey/dev/maria-5.3-subqueries-r7/
by Sergey Petrunya 15 Mar '10

15 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7/ ------------------------------------------------------------ revno: 2777 revision-id: psergey(a)askmonty.org-20100315060659-0spqc4jdav12ja2u parent: psergey(a)askmonty.org-20100314182543-4t3ehit7df20adu8 committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7 timestamp: Mon 2010-03-15 09:06:59 +0300 message: Update test results for the previous push === modified file 'mysql-test/r/type_datetime.result' --- a/mysql-test/r/type_datetime.result 2010-02-11 21:59:32 +0000 +++ b/mysql-test/r/type_datetime.result 2010-03-15 06:06:59 +0000 @@ -516,7 +516,7 @@ 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables Warnings: Note 1276 Field or reference 'test.t1.cur_date' of SELECT #2 was resolved in SELECT #1 -Note 1003 select '1' AS `id`,'2007-04-25 18:30:22' AS `cur_date` from `test`.`t1` `x1` join `test`.`t1` where (('2007-04-25 18:30:22' = 0)) +Note 1003 select '1' AS `id`,'2007-04-25 18:30:22' AS `cur_date` from `test`.`t1` semi join (`test`.`t1` `x1`) where (('2007-04-25 18:30:22' = 0)) select * from t1 where id in (select id from t1 as x1 where (t1.cur_date is null)); id cur_date @@ -527,7 +527,7 @@ 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables Warnings: Note 1276 Field or reference 'test.t2.cur_date' of SELECT #2 was resolved in SELECT #1 -Note 1003 select '1' AS `id`,'2007-04-25' AS `cur_date` from `test`.`t2` `x1` join `test`.`t2` where (('2007-04-25' = 0)) +Note 1003 select '1' AS `id`,'2007-04-25' AS `cur_date` from `test`.`t2` semi join (`test`.`t2` `x1`) where (('2007-04-25' = 0)) select * from t2 where id in (select id from t2 as x1 where (t2.cur_date is null)); id cur_date

1 0

[Maria-developers] Rev 2776: Merge in file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/
by Sergey Petrunya 14 Mar '10

14 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/ ------------------------------------------------------------ revno: 2776 [merge] revision-id: psergey(a)askmonty.org-20100314182543-4t3ehit7df20adu8 parent: psergey(a)askmonty.org-20100314175549-0gcze3pxaudgapxh parent: psergey(a)askmonty.org-20100313211106-5xyfyl02gfenbi7f committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7-rel timestamp: Sun 2010-03-14 21:25:43 +0300 message: Merge modified: mysql-test/r/subselect_mat.result subselect_mat.result-20100117143924-r0jv32dj80dg3b5h-1 mysql-test/r/subselect_sj.result subselect_sj.result-20100117143926-nrop4ku355g3kv8b-1 mysql-test/r/subselect_sj_jcl6.result subselect_sj_jcl6.re-20100117143928-7vzk51yaf29cdavp-1 mysql-test/t/subselect_mat.test subselect_mat.test-20100117143929-iif102ysgna1tyj0-1 mysql-test/t/subselect_sj.test subselect_sj.test-20100117143931-qp396ufpe3k0scre-1 sql/item.cc sp1f-item.cc-19700101030959-u7hxqopwpfly4kf5ctlyk2dvrq4l3dhn sql/item_cmpfunc.cc sp1f-item_cmpfunc.cc-19700101030959-hrk7pi2n6qpwxauufnkizirsoucdcx2e sql/item_cmpfunc.h sp1f-item_cmpfunc.h-19700101030959-pcvbjplo4e4ng7ibynfhcd6pjyem57gr sql/opt_subselect.cc opt_subselect.cc-20100215190428-nekkl8wisp0k6nlk-1 sql/sql_select.cc sp1f-sql_select.cc-19700101030959-egb7whpkh76zzvikycs5nsnuviu4fdlb sql/sql_select.h sp1f-sql_select.h-19700101030959-oqegfxr76xlgmrzd6qlevonoibfnwzoz === modified file 'mysql-test/r/subselect_mat.result' --- a/mysql-test/r/subselect_mat.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/subselect_mat.result 2010-03-13 21:11:06 +0000 @@ -583,7 +583,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a1` in (select 1 AS `Not_used` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`)))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`)))) select left(a1,7), left(a2,7) from t1_16 where a1 in (select b1 from t2_16 where b1 > '0'); @@ -597,7 +597,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`) in (select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),<exists>(select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) select left(a1,7), left(a2,7) from t1_16 where (a1,a2) in (select b1, b2 from t2_16 where b1 > '0'); @@ -625,7 +625,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using filesort Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a1` in (select group_concat(`test`.`t2_16`.`b1` separator ',') AS `group_concat(b1)` from `test`.`t2_16` group by `test`.`t2_16`.`b2` having (<cache>(`test`.`t1_16`.`a1`) = <ref_null_helper>(group_concat(`test`.`t2_16`.`b1` separator ','))))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,<exists>(select group_concat(`test`.`t2_16`.`b1` separator ',') AS `group_concat(b1)` from `test`.`t2_16` group by `test`.`t2_16`.`b2` having (<cache>(`test`.`t1_16`.`a1`) = <ref_null_helper>(group_concat(`test`.`t2_16`.`b1` separator ','))))) select left(a1,7), left(a2,7) from t1_16 where a1 in (select group_concat(b1) from t2_16 group by b2); @@ -662,7 +662,7 @@ 3 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 5 100.00 Using where; Using join buffer 4 SUBQUERY t3 ALL NULL NULL NULL NULL 4 100.00 Using where Warnings: -Note 1003 select `test`.`t1`.`a1` AS `a1`,`test`.`t1`.`a2` AS `a2` from `test`.`t1` where <in_optimizer>(concat(`test`.`t1`.`a1`,'x'),<exists>(select 1 AS `Not_used` from `test`.`t1_16` where (<in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`) in (select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` join `test`.`t2` where ((`test`.`t2`.`b2` = substr(`test`.`t2_16`.`b2`,1,6)) and <in_optimizer>(`test`.`t2`.`b1`,`test`.`t2`.`b1` in ( <materialize> (select `test`.`t3`.`c1` AS `c1` from `test`.`t3` where (`test`.`t3`.`c2` > '0') ), <primary_index_lookup>(`test`.`t2`.`b1` in <temporary table> on distinct_key where ((`test`.`t2`.`b1` = `materialized subselect`.`c1`))))) and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) and (<cache>(concat(`test`.`t1`.`a1`,'x')) = left(`test`.`t1_16`.`a1`,8))))) +Note 1003 select `test`.`t1`.`a1` AS `a1`,`test`.`t1`.`a2` AS `a2` from `test`.`t1` where <in_optimizer>(concat(`test`.`t1`.`a1`,'x'),<exists>(select 1 AS `Not_used` from `test`.`t1_16` where (<in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),<exists>(select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` join `test`.`t2` where ((`test`.`t2`.`b2` = substr(`test`.`t2_16`.`b2`,1,6)) and <in_optimizer>(`test`.`t2`.`b1`,`test`.`t2`.`b1` in ( <materialize> (select `test`.`t3`.`c1` AS `c1` from `test`.`t3` where (`test`.`t3`.`c2` > '0') ), <primary_index_lookup>(`test`.`t2`.`b1` in <temporary table> on distinct_key where ((`test`.`t2`.`b1` = `materialized subselect`.`c1`))))) and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) and (<cache>(concat(`test`.`t1`.`a1`,'x')) = left(`test`.`t1_16`.`a1`,8))))) drop table t1_16, t2_16, t3_16; set @blob_len = 512; set @suffix_len = @blob_len - @prefix_len; @@ -696,7 +696,7 @@ 1 PRIMARY t1_512 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_512 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>(`test`.`t1_512`.`a1`,`test`.`t1_512`.`a1` in (select 1 AS `Not_used` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`)))) +Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>(`test`.`t1_512`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`)))) select left(a1,7), left(a2,7) from t1_512 where a1 in (select b1 from t2_512 where b1 > '0'); @@ -710,7 +710,7 @@ 1 PRIMARY t1_512 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_512 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>((`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`),(`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`) in (select `test`.`t2_512`.`b1` AS `b1`,`test`.`t2_512`.`b2` AS `b2` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`) and (<cache>(`test`.`t1_512`.`a2`) = `test`.`t2_512`.`b2`)))) +Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>((`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`),<exists>(select `test`.`t2_512`.`b1` AS `b1`,`test`.`t2_512`.`b2` AS `b2` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`) and (<cache>(`test`.`t1_512`.`a2`) = `test`.`t2_512`.`b2`)))) select left(a1,7), left(a2,7) from t1_512 where (a1,a2) in (select b1, b2 from t2_512 where b1 > '0'); @@ -789,7 +789,7 @@ 1 PRIMARY t1_1024 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1024 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>(`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a1` in (select 1 AS `Not_used` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`)))) +Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>(`test`.`t1_1024`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`)))) select left(a1,7), left(a2,7) from t1_1024 where a1 in (select b1 from t2_1024 where b1 > '0'); @@ -803,7 +803,7 @@ 1 PRIMARY t1_1024 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1024 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>((`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`),(`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`) in (select `test`.`t2_1024`.`b1` AS `b1`,`test`.`t2_1024`.`b2` AS `b2` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`) and (<cache>(`test`.`t1_1024`.`a2`) = `test`.`t2_1024`.`b2`)))) +Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>((`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`),<exists>(select `test`.`t2_1024`.`b1` AS `b1`,`test`.`t2_1024`.`b2` AS `b2` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`) and (<cache>(`test`.`t1_1024`.`a2`) = `test`.`t2_1024`.`b2`)))) select left(a1,7), left(a2,7) from t1_1024 where (a1,a2) in (select b1, b2 from t2_1024 where b1 > '0'); @@ -882,7 +882,7 @@ 1 PRIMARY t1_1025 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1025 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>(`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a1` in (select 1 AS `Not_used` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`)))) +Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>(`test`.`t1_1025`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`)))) select left(a1,7), left(a2,7) from t1_1025 where a1 in (select b1 from t2_1025 where b1 > '0'); @@ -896,7 +896,7 @@ 1 PRIMARY t1_1025 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1025 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>((`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`),(`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`) in (select `test`.`t2_1025`.`b1` AS `b1`,`test`.`t2_1025`.`b2` AS `b2` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`) and (<cache>(`test`.`t1_1025`.`a2`) = `test`.`t2_1025`.`b2`)))) +Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>((`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`),<exists>(select `test`.`t2_1025`.`b1` AS `b1`,`test`.`t2_1025`.`b2` AS `b2` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`) and (<cache>(`test`.`t1_1025`.`a2`) = `test`.`t2_1025`.`b2`)))) select left(a1,7), left(a2,7) from t1_1025 where (a1,a2) in (select b1, b2 from t2_1025 where b1 > '0'); @@ -982,7 +982,7 @@ 1 PRIMARY t1bb ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2bb ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select conv(`test`.`t1bb`.`a1`,10,2) AS `bin(a1)`,`test`.`t1bb`.`a2` AS `a2` from `test`.`t1bb` where <in_optimizer>((`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`),(`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`) in (select `test`.`t2bb`.`b1` AS `b1`,`test`.`t2bb`.`b2` AS `b2` from `test`.`t2bb` where ((<cache>(`test`.`t1bb`.`a1`) = `test`.`t2bb`.`b1`) and (<cache>(`test`.`t1bb`.`a2`) = `test`.`t2bb`.`b2`)))) +Note 1003 select conv(`test`.`t1bb`.`a1`,10,2) AS `bin(a1)`,`test`.`t1bb`.`a2` AS `a2` from `test`.`t1bb` where <in_optimizer>((`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`),<exists>(select `test`.`t2bb`.`b1` AS `b1`,`test`.`t2bb`.`b2` AS `b2` from `test`.`t2bb` where ((<cache>(`test`.`t1bb`.`a1`) = `test`.`t2bb`.`b1`) and (<cache>(`test`.`t1bb`.`a2`) = `test`.`t2bb`.`b2`)))) select bin(a1), a2 from t1bb where (a1, a2) in (select b1, b2 from t2bb); @@ -1219,3 +1219,28 @@ pk 2 DROP TABLE t1, t2; +# +# BUG#50019: Wrong result for IN-subquery with materialization +# +create table t1(i int); +insert into t1 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t2(i int); +insert into t2 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t3(i int); +insert into t3 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +i +1 +2 +3 +4 +set @save_optimizer_switch=@@optimizer_switch; +set session optimizer_switch='materialization=off'; +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +i +1 +2 +3 +4 +set session optimizer_switch=@save_optimizer_switch; +drop table t1, t2, t3; === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-14 17:54:12 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-14 18:25:43 +0000 @@ -825,6 +825,127 @@ 2 drop table t1, t2, t3; # +# Bug#48213 Materialized subselect crashes if using GEOMETRY type +# +CREATE TABLE t1 ( +pk int, +a varchar(1), +b varchar(4), +c tinyblob, +d blob, +e mediumblob, +f longblob, +g tinytext, +h text, +i mediumtext, +j longtext, +k geometry, +PRIMARY KEY (pk) +); +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using MRR; Materialize +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`c` = `test`.`t1`.`c`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`d` = `test`.`t1`.`d`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`e` = `test`.`t1`.`e`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`f` = `test`.`t1`.`f`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`g` = `test`.`t1`.`g`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`h` = `test`.`t1`.`h`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`i` = `test`.`t1`.`i`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`j` = `test`.`t1`.`j`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`k` = `test`.`t1`.`k`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +pk +1 +2 +DROP TABLE t1, t2; +# End of Bug#48213 +# # Bug#49198 Wrong result for second call of procedure # with view in subselect. # @@ -872,6 +993,42 @@ DROP VIEW v2, v3; # End of Bug#49198 # +# Bug#45174: Incorrectly applied equality propagation caused wrong +# result on a query with a materialized semi-join. +# +CREATE TABLE `t1` ( +`pk` int(11) NOT NULL AUTO_INCREMENT, +`varchar_key` varchar(1) NOT NULL, +`varchar_nokey` varchar(1) NOT NULL, +PRIMARY KEY (`pk`), +KEY `varchar_key` (`varchar_key`) +); +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); +CREATE TABLE `t2` ( +`varchar_nokey` varchar(1) NOT NULL +); +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t2 ALL NULL NULL NULL NULL 18 100.00 +1 PRIMARY t1 ALL varchar_key NULL NULL NULL 15 100.00 Using where; Materialize +Warnings: +Note 1003 select `test`.`t2`.`varchar_nokey` AS `varchar_nokey` from `test`.`t2` semi join (`test`.`t1`) where ((`test`.`t1`.`varchar_nokey` = `test`.`t1`.`varchar_key`) and ((`test`.`t1`.`varchar_nokey` < 'n') xor `test`.`t1`.`pk`)) +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +varchar_nokey +DROP TABLE t1, t2; +# End of the test for bug#45174. +# # BUG#43768: Prepared query with nested subqueries core dumps on second execution # create table t1 ( === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-14 17:54:12 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-14 18:25:43 +0000 @@ -829,6 +829,127 @@ 2 drop table t1, t2, t3; # +# Bug#48213 Materialized subselect crashes if using GEOMETRY type +# +CREATE TABLE t1 ( +pk int, +a varchar(1), +b varchar(4), +c tinyblob, +d blob, +e mediumblob, +f longblob, +g tinytext, +h text, +i mediumtext, +j longtext, +k geometry, +PRIMARY KEY (pk) +); +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using MRR; Materialize +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`c` = `test`.`t1`.`c`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`d` = `test`.`t1`.`d`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`e` = `test`.`t1`.`e`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`f` = `test`.`t1`.`f`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`g` = `test`.`t1`.`g`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`h` = `test`.`t1`.`h`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`i` = `test`.`t1`.`i`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`j` = `test`.`t1`.`j`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`k` = `test`.`t1`.`k`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +pk +1 +2 +DROP TABLE t1, t2; +# End of Bug#48213 +# # Bug#49198 Wrong result for second call of procedure # with view in subselect. # @@ -876,6 +997,42 @@ DROP VIEW v2, v3; # End of Bug#49198 # +# Bug#45174: Incorrectly applied equality propagation caused wrong +# result on a query with a materialized semi-join. +# +CREATE TABLE `t1` ( +`pk` int(11) NOT NULL AUTO_INCREMENT, +`varchar_key` varchar(1) NOT NULL, +`varchar_nokey` varchar(1) NOT NULL, +PRIMARY KEY (`pk`), +KEY `varchar_key` (`varchar_key`) +); +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); +CREATE TABLE `t2` ( +`varchar_nokey` varchar(1) NOT NULL +); +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t2 ALL NULL NULL NULL NULL 18 100.00 +1 PRIMARY t1 ALL varchar_key NULL NULL NULL 15 100.00 Using where; Materialize +Warnings: +Note 1003 select `test`.`t2`.`varchar_nokey` AS `varchar_nokey` from `test`.`t2` semi join (`test`.`t1`) where ((`test`.`t1`.`varchar_nokey` = `test`.`t1`.`varchar_key`) and ((`test`.`t1`.`varchar_nokey` < 'n') xor `test`.`t1`.`pk`)) +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +varchar_nokey +DROP TABLE t1, t2; +# End of the test for bug#45174. +# # BUG#43768: Prepared query with nested subqueries core dumps on second execution # create table t1 ( === modified file 'mysql-test/t/subselect_mat.test' --- a/mysql-test/t/subselect_mat.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect_mat.test 2010-03-13 20:04:52 +0000 @@ -889,3 +889,19 @@ SELECT pk FROM t1 WHERE (b,c,d) IN (SELECT b,c,d FROM t2 WHERE pk > 0); DROP TABLE t1, t2; +--echo # +--echo # BUG#50019: Wrong result for IN-subquery with materialization +--echo # +create table t1(i int); +insert into t1 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t2(i int); +insert into t2 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t3(i int); +insert into t3 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +set @save_optimizer_switch=@@optimizer_switch; +set session optimizer_switch='materialization=off'; +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +set session optimizer_switch=@save_optimizer_switch; +drop table t1, t2, t3; + === modified file 'mysql-test/t/subselect_sj.test' --- a/mysql-test/t/subselect_sj.test 2010-03-14 17:54:12 +0000 +++ b/mysql-test/t/subselect_sj.test 2010-03-14 18:25:43 +0000 @@ -729,6 +729,86 @@ drop table t1, t2, t3; --echo # +--echo # Bug#48213 Materialized subselect crashes if using GEOMETRY type +--echo # + +CREATE TABLE t1 ( + pk int, + a varchar(1), + b varchar(4), + c tinyblob, + d blob, + e mediumblob, + f longblob, + g tinytext, + h text, + i mediumtext, + j longtext, + k geometry, + PRIMARY KEY (pk) +); + +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); + +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); + +# Test that materialization is skipped for semijoins where materialized +# table would contain GEOMETRY or different kinds of BLOB/TEXT columns +let $query= +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +DROP TABLE t1, t2; +--echo # End of Bug#48213 + +--echo # --echo # Bug#49198 Wrong result for second call of procedure --echo # with view in subselect. --echo # @@ -772,6 +852,44 @@ --echo # End of Bug#49198 --echo # +--echo # Bug#45174: Incorrectly applied equality propagation caused wrong +--echo # result on a query with a materialized semi-join. +--echo # + +CREATE TABLE `t1` ( + `pk` int(11) NOT NULL AUTO_INCREMENT, + `varchar_key` varchar(1) NOT NULL, + `varchar_nokey` varchar(1) NOT NULL, + PRIMARY KEY (`pk`), + KEY `varchar_key` (`varchar_key`) +); + +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); + +CREATE TABLE `t2` ( + `varchar_nokey` varchar(1) NOT NULL +); + +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); + +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; + +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; + +DROP TABLE t1, t2; + +--echo # End of the test for bug#45174. +--echo # --echo # BUG#43768: Prepared query with nested subqueries core dumps on second execution --echo # create table t1 ( === modified file 'sql/item.cc' --- a/sql/item.cc 2010-02-24 11:33:42 +0000 +++ b/sql/item.cc 2010-03-13 20:04:52 +0000 @@ -4761,7 +4761,7 @@ return this; return const_item; } - Item_field *subst= item_equal->get_first(); + Item_field *subst= item_equal->get_first(this); if (subst && field->table != subst->field->table && !field->eq(subst->field)) return subst; } === modified file 'sql/item_cmpfunc.cc' --- a/sql/item_cmpfunc.cc 2010-02-17 10:05:27 +0000 +++ b/sql/item_cmpfunc.cc 2010-03-13 20:04:52 +0000 @@ -5369,7 +5369,7 @@ void Item_equal::fix_length_and_dec() { - Item *item= get_first(); + Item *item= get_first(NULL); eval_item= cmp_item::get_comparator(item->result_type(), item->collation.collation); } @@ -5432,3 +5432,128 @@ str->append(')'); } + +/* + @brief Get the first equal field of multiple equality. + @param[in] field the field to get equal field to + + @details Get the first field of multiple equality that is equal to the + given field. In order to make semi-join materialization strategy work + correctly we can't propagate equal fields from upper select to a + materialized semi-join. + Thus the fields is returned according to following rules: + + 1) If the given field belongs to a semi-join then the first field in + multiple equality which belong to the same semi-join is returned. + Otherwise NULL is returned. + 2) If the given field doesn't belong to a semi-join then + the first field in the multiple equality that doesn't belong to any + semi-join is returned. + If all fields in the equality are belong to semi-join(s) then NULL + is returned. + 3) If no field is given then the first field in the multiple equality + is returned without regarding whether it belongs to a semi-join or not. + + @retval Found first field in the multiple equality. + @retval 0 if no field found. +*/ + +Item_field* Item_equal::get_first(Item_field *field) +{ + List_iterator<Item_field> it(fields); + Item_field *item; + JOIN_TAB *field_tab; + + if (!field) + return fields.head(); + + /* + Of all equal fields, return the first one we can use. Normally, this is the + field which belongs to the table that is the first in the join order. + + There is one exception to this: When semi-join materialization strategy is + used, and the given field belongs to a table within the semi-join nest, we + must pick the first field in the semi-join nest. + + Example: suppose we have a join order: + + ot1 ot2 SJ-Mat(it1 it2 it3) ot3 + + and equality ot2.col = it1.col = it2.col + If we're looking for best substitute for 'it2.col', we should pick it1.col + and not ot2.col. + + eliminate_item_equal() also has code that deals with equality substitution + in presense of SJM nests. + */ + + field_tab= field->field->table->reginfo.join_tab; + + TABLE_LIST *emb_nest= field->field->table->pos_in_table_list->embedding; + + if (emb_nest && emb_nest->sj_mat_info && emb_nest->sj_mat_info->is_used) + { + /* + It's a field from an materialized semi-join. We can substitute it only + for a field from the same semi-join. + */ + JOIN_TAB *first; + JOIN *join= field_tab->join; + uint tab_idx= field_tab - field_tab->join->join_tab; + + /* Find the first table of this semi-join nest */ + for (uint i= tab_idx; i != join->const_tables; i--) + { + if (join->join_tab[i].table->map & emb_nest->sj_inner_tables) + first= join->join_tab + i; + else + // Found first tab that doesn't belong to current SJ. + break; + } + /* Find an item to substitute for. */ + while ((item= it++)) + { + if (item->field->table->reginfo.join_tab >= first) + { + /* + If we found given field then return NULL to avoid unnecessary + substitution. + */ + return (item != field) ? item : NULL; + } + } + } + else + { +#if 0 + /* + The field is not in SJ-Materialization nest. We must return the first + field that's not embedded in a SJ-Materialization nest. + Example: suppose we have a join order: + + SJ-Mat(it1 it2) ot1 ot2 + + and equality ot2.col = ot1.col = it2.col + If we're looking for best substitute for 'ot2.col', we should pick ot1.col + and not it2.col, because when we run a join between ot1 and ot2 + execution of SJ-Mat(...) has already finished and we can't rely on the + value of it*.*. + psergey-fix-fix: ^^ THAT IS INCORRECT ^^. Pick the first, whatever that + is. + */ + while ((item= it++)) + { + TABLE_LIST *emb_nest= item->field->table->pos_in_table_list->embedding; + if (!emb_nest || !emb_nest->sj_mat_info || + !emb_nest->sj_mat_info->is_used) + { + return item; + } + } +#endif + return fields.head(); + } + // Shouldn't get here. + DBUG_ASSERT(0); + return NULL; +} === modified file 'sql/item_cmpfunc.h' --- a/sql/item_cmpfunc.h 2010-02-17 10:05:27 +0000 +++ b/sql/item_cmpfunc.h 2010-03-13 20:04:52 +0000 @@ -1589,7 +1589,7 @@ void add(Item_field *f); uint members(); bool contains(Field *field); - Item_field* get_first() { return fields.head(); } + Item_field* get_first(Item_field *field); uint n_fields() { return fields.elements; } void merge(Item_equal *item); void update_const(); === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-14 17:54:12 +0000 +++ b/sql/opt_subselect.cc 2010-03-14 18:25:43 +0000 @@ -322,7 +322,13 @@ default: ;/* suitable for materialization */ } + + // Materialization does not work with BLOB columns + if (inner->field_type() == MYSQL_TYPE_BLOB || + inner->field_type() == MYSQL_TYPE_GEOMETRY) + DBUG_RETURN(FALSE); } + in_subs->types_allow_materialization= TRUE; in_subs->sjm_scan_allowed= all_are_fields; DBUG_PRINT("info",("subquery_types_allow_materialization: ok, allowed")); @@ -2181,6 +2187,8 @@ if (tablenr != first) pos->sj_strategy= SJ_OPT_NONE; remaining_tables |= s->table->map; + //s->sj_strategy= pos->sj_strategy; + join->join_tab[first].sj_strategy= join->best_positions[first].sj_strategy; } } === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-14 17:54:12 +0000 +++ b/sql/sql_select.cc 2010-03-14 18:25:43 +0000 @@ -8869,6 +8869,15 @@ } +static TABLE_LIST* embedding_sjm(Item_field *item_field) +{ + TABLE_LIST *nest= item_field->field->table->pos_in_table_list->embedding; + if (nest && nest->sj_mat_info && nest->sj_mat_info->is_used) + return nest; + else + return NULL; +} + /** Generate minimal set of simple equalities equivalent to a multiple equality. @@ -8902,6 +8911,23 @@ So only t1.a=t3.c should be left in the lower level. If cond is equal to 0, then not more then one equality is generated and a pointer to it is returned as the result of the function. + + Equality substutution and semi-join materialization nests: + + In case join order looks like this: + + outer_tbl1 outer_tbl2 SJM (inner_tbl1 inner_tbl2) outer_tbl3 + + We must not construct equalities like + + outer_tbl1.col = inner_tbl1.col + + because they would get attached to inner_tbl1 and will get evaluated + during materialization phase, when we don't have current value of + outer_tbl1.col. + + Item_equal::get_first() also takes similar measures for dealing with + equality substitution in presense of SJM nests. @return - The condition with generated simple equalities or @@ -8919,18 +8945,44 @@ Item *item_const= item_equal->get_const(); Item_equal_iterator it(*item_equal); Item *head; + TABLE_LIST *current_sjm= NULL; + Item *current_sjm_head= NULL; + + /* + Pick the "head" item: the constant one or the first in the join order + that's not inside some SJM nest. + */ if (item_const) head= item_const; else { - head= item_equal->get_first(); + TABLE_LIST *emb_nest; + Item_field *item_field; + head= item_field= item_equal->get_first(NULL); it++; + if ((emb_nest= embedding_sjm(item_field))) + { + current_sjm= emb_nest; + current_sjm_head= head; + } } + Item_field *item_field; + /* + For each other item, generate "item=head" equality (except the tables that + are within SJ-Materialization nests, for those "head" is defined + differently) + */ while ((item_field= it++)) { Item_equal *upper= item_field->find_item_equal(upper_levels); Item_field *item= item_field; + TABLE_LIST *field_sjm= embedding_sjm(item_field); + + /* + Check if "item_field=head" equality is already guaranteed to be true + on upper AND-levels. + */ if (upper) { if (item_const && upper->get_const()) @@ -8945,65 +8997,29 @@ } } } - if (item == item_field) + + bool produce_equality= test(item == item_field); + if (!item_const && field_sjm && field_sjm != current_sjm) + { + /* Entering an SJM nest */ + current_sjm_head= item_field; + if (!field_sjm->sj_mat_info->is_sj_scan) + produce_equality= FALSE; + } + + if (produce_equality) { if (eq_item) eq_list.push_back(eq_item); - /* - item_field might refer to a table that is within a semi-join - materialization nest. In that case, the join order looks like this: - - outer_tbl1 outer_tbl2 SJM (inner_tbl1 inner_tbl2) outer_tbl3 - - We must not construct equalities like - - outer_tbl1.col = inner_tbl1.col - - because they would get attached to inner_tbl1 and will get evaluated - during materialization phase, when we don't have current value of - outer_tbl1.col. - */ - TABLE_LIST *emb_nest= - item_field->field->table->pos_in_table_list->embedding; - if (!item_const && emb_nest && emb_nest->sj_mat_info && - emb_nest->sj_mat_info->is_used) - { - /* - Find the first equal expression that refers to a table that is - within the semijoin nest. If we can't find it, do nothing - */ - List_iterator<Item_field> fit(item_equal->fields); - Item_field *head_in_sjm; - bool found= FALSE; - while ((head_in_sjm= fit++)) - { - if (head_in_sjm->used_tables() & emb_nest->sj_inner_tables) - { - if (head_in_sjm == item_field) - { - /* This is the first table inside the semi-join*/ - eq_item= new Item_func_eq(item_field, head); - /* Tell make_cond_for_table don't use this. */ - eq_item->marker=3; - } - else - { - eq_item= new Item_func_eq(item_field, head_in_sjm); - found= TRUE; - } - break; - } - } - if (!found) - continue; - } - else - eq_item= new Item_func_eq(item_field, head); + + eq_item= new Item_func_eq(item_field, current_sjm? current_sjm_head: head); + if (!eq_item) return 0; eq_item->set_cmp_func(); eq_item->quick_fix_field(); } + current_sjm= field_sjm; } if (!cond && !eq_list.head()) === modified file 'sql/sql_select.h' --- a/sql/sql_select.h 2010-03-05 18:54:48 +0000 +++ b/sql/sql_select.h 2010-03-13 20:04:52 +0000 @@ -279,6 +279,13 @@ /* NestedOuterJoins: Bitmap of nested joins this table is part of */ nested_join_map embedding_map; + /* + Semi-join strategy to be used for this join table. This is a copy of + POSITION::sj_strategy field. This field is set up by the + fix_semijion_strategies_for_picked_join_order. + */ + uint sj_strategy; + void cleanup(); inline bool is_using_loose_index_scan() {

1 0

[Maria-developers] Rev 2775: Fix support-files/build-tags to work with recent versions of bazaar. in file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/
by Sergey Petrunya 14 Mar '10

14 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/ ------------------------------------------------------------ revno: 2775 revision-id: psergey(a)askmonty.org-20100314175549-0gcze3pxaudgapxh parent: psergey(a)askmonty.org-20100314175412-umtxuabkn4txl1yd committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7-rel timestamp: Sun 2010-03-14 20:55:49 +0300 message: Fix support-files/build-tags to work with recent versions of bazaar. === modified file 'support-files/build-tags' --- a/support-files/build-tags 2009-12-15 07:16:46 +0000 +++ b/support-files/build-tags 2010-03-14 17:55:49 +0000 @@ -4,7 +4,7 @@ filter='\.cc$\|\.c$\|\.h$\|\.yy$' list="find . -type f" -bzr root >/dev/null 2>/dev/null && list="bzr ls --from-root --kind=file --versioned" +bzr root >/dev/null 2>/dev/null && list="bzr ls --from-root -R --kind=file --versioned" $list |grep $filter |while read f; do

1 0

[Maria-developers] Rev 2774: BUG#43768: Prepared query with nested subqueries core dumps on second execution in file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/
by Sergey Petrunya 14 Mar '10

14 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7-rel/ ------------------------------------------------------------ revno: 2774 revision-id: psergey(a)askmonty.org-20100314175412-umtxuabkn4txl1yd parent: psergey(a)askmonty.org-20100307154145-ksby2b1l0sqm1xne committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7-rel timestamp: Sun 2010-03-14 20:54:12 +0300 message: BUG#43768: Prepared query with nested subqueries core dumps on second execution Fix two problems: 1. Let optimize_semijoin_nests() reset sj_nest->sjmat_info irrespectively of value of optimizer_flag. We need this in case somebody has turned optimization off between reexecutions of the same statement. 2. Do not pull out constant tables out of semi-join nests. The problem is that pullout operation is not undoable, and if a table is constant because it is 1/0-row table it may cease to be constant on the next execution. Note that tables that are constant because of possible eq_ref(const) access will still be pulled out as they are considered functionally-dependent. === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-02-24 11:33:42 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-14 17:54:12 +0000 @@ -1,4 +1,4 @@ -drop table if exists t0, t1, t2, t10, t11, t12; +drop table if exists t0, t1, t2, t3, t4, t10, t11, t12; create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1(a int, b int); @@ -871,3 +871,54 @@ DROP TABLE t1, t2, t3; DROP VIEW v2, v3; # End of Bug#49198 +# +# BUG#43768: Prepared query with nested subqueries core dumps on second execution +# +create table t1 ( +id int(11) unsigned not null primary key auto_increment, +partner_id varchar(35) not null, +t1_status_id int(10) unsigned +); +insert into t1 values ("1", "partner1", "10"), ("2", "partner2", "10"), +("3", "partner3", "10"), ("4", "partner4", "10"); +create table t2 ( +id int(11) unsigned not null default '0', +t1_line_id int(11) unsigned not null default '0', +article_id varchar(20), +sequence int(11) not null default '0', +primary key (id,t1_line_id) +); +insert into t2 values ("1", "1", "sup", "0"), ("2", "1", "sup", "1"), +("2", "2", "sup", "2"), ("2", "3", "sup", "3"), +("2", "4", "imp", "4"), ("3", "1", "sup", "0"), +("4", "1", "sup", "0"); +create table t3 ( +id int(11) not null default '0', +preceeding_id int(11) not null default '0', +primary key (id,preceeding_id) +); +create table t4 ( +user_id varchar(50) not null, +article_id varchar(20) not null, +primary key (user_id,article_id) +); +insert into t4 values("nicke", "imp"); +prepare stmt from +'select t1.partner_id +from t1 +where + t1.id in ( + select pl_inner.id + from t2 as pl_inner + where pl_inner.article_id in ( + select t4.article_id from t4 + where t4.user_id = \'nicke\' + ) + )'; +execute stmt; +partner_id +partner2 +execute stmt; +partner_id +partner2 +drop table t1,t2,t3,t4; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-07 15:41:45 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-14 17:54:12 +0000 @@ -2,7 +2,7 @@ show variables like 'join_cache_level'; Variable_name Value join_cache_level 6 -drop table if exists t0, t1, t2, t10, t11, t12; +drop table if exists t0, t1, t2, t3, t4, t10, t11, t12; create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1(a int, b int); @@ -876,6 +876,57 @@ DROP VIEW v2, v3; # End of Bug#49198 # +# BUG#43768: Prepared query with nested subqueries core dumps on second execution +# +create table t1 ( +id int(11) unsigned not null primary key auto_increment, +partner_id varchar(35) not null, +t1_status_id int(10) unsigned +); +insert into t1 values ("1", "partner1", "10"), ("2", "partner2", "10"), +("3", "partner3", "10"), ("4", "partner4", "10"); +create table t2 ( +id int(11) unsigned not null default '0', +t1_line_id int(11) unsigned not null default '0', +article_id varchar(20), +sequence int(11) not null default '0', +primary key (id,t1_line_id) +); +insert into t2 values ("1", "1", "sup", "0"), ("2", "1", "sup", "1"), +("2", "2", "sup", "2"), ("2", "3", "sup", "3"), +("2", "4", "imp", "4"), ("3", "1", "sup", "0"), +("4", "1", "sup", "0"); +create table t3 ( +id int(11) not null default '0', +preceeding_id int(11) not null default '0', +primary key (id,preceeding_id) +); +create table t4 ( +user_id varchar(50) not null, +article_id varchar(20) not null, +primary key (user_id,article_id) +); +insert into t4 values("nicke", "imp"); +prepare stmt from +'select t1.partner_id +from t1 +where + t1.id in ( + select pl_inner.id + from t2 as pl_inner + where pl_inner.article_id in ( + select t4.article_id from t4 + where t4.user_id = \'nicke\' + ) + )'; +execute stmt; +partner_id +partner2 +execute stmt; +partner_id +partner2 +drop table t1,t2,t3,t4; +# # BUG#49129: Wrong result with IN-subquery with join_cache_level=6 and firstmatch=off # CREATE TABLE t0 (a INT); === modified file 'mysql-test/t/subselect_sj.test' --- a/mysql-test/t/subselect_sj.test 2010-02-24 11:33:42 +0000 +++ b/mysql-test/t/subselect_sj.test 2010-03-14 17:54:12 +0000 @@ -2,7 +2,7 @@ # Nested Loops semi-join subquery evaluation tests # --disable_warnings -drop table if exists t0, t1, t2, t10, t11, t12; +drop table if exists t0, t1, t2, t3, t4, t10, t11, t12; --enable_warnings # @@ -770,3 +770,60 @@ DROP VIEW v2, v3; --echo # End of Bug#49198 + +--echo # +--echo # BUG#43768: Prepared query with nested subqueries core dumps on second execution +--echo # +create table t1 ( + id int(11) unsigned not null primary key auto_increment, + partner_id varchar(35) not null, + t1_status_id int(10) unsigned +); + +insert into t1 values ("1", "partner1", "10"), ("2", "partner2", "10"), + ("3", "partner3", "10"), ("4", "partner4", "10"); + +create table t2 ( + id int(11) unsigned not null default '0', + t1_line_id int(11) unsigned not null default '0', + article_id varchar(20), + sequence int(11) not null default '0', + primary key (id,t1_line_id) +); + +insert into t2 values ("1", "1", "sup", "0"), ("2", "1", "sup", "1"), + ("2", "2", "sup", "2"), ("2", "3", "sup", "3"), + ("2", "4", "imp", "4"), ("3", "1", "sup", "0"), + ("4", "1", "sup", "0"); +create table t3 ( + id int(11) not null default '0', + preceeding_id int(11) not null default '0', + primary key (id,preceeding_id) +); + +create table t4 ( + user_id varchar(50) not null, + article_id varchar(20) not null, + primary key (user_id,article_id) +); + +insert into t4 values("nicke", "imp"); +prepare stmt from +'select t1.partner_id +from t1 +where + t1.id in ( + select pl_inner.id + from t2 as pl_inner + where pl_inner.article_id in ( + select t4.article_id from t4 + where t4.user_id = \'nicke\' + ) + )'; + +execute stmt; +execute stmt; +drop table t1,t2,t3,t4; + + + === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-07 15:41:45 +0000 +++ b/sql/opt_subselect.cc 2010-03-14 17:54:12 +0000 @@ -963,7 +963,6 @@ { /* Action #1: Mark the constant tables to be pulled out */ table_map pulled_tables= 0; - List_iterator<TABLE_LIST> child_li(sj_nest->nested_join->join_list); TABLE_LIST *tbl; while ((tbl= child_li++)) @@ -971,12 +970,34 @@ if (tbl->table) { tbl->table->reginfo.join_tab->emb_sj_nest= sj_nest; +#if 0 + /* + Do not pull out tables because they are constant. This operation has + a problem: + - Some constant tables may become/cease to be constant across PS + re-executions + - Contrary to our initial assumption, it turned out that table pullout + operation is not easily undoable. + + The solution is to leave constant tables where they are. This will + affect only constant tables that are 1-row or empty, tables that are + constant because they are accessed via eq_ref(const) access will + still be pulled out as functionally-dependent. + + This will cause us to miss the chance to flatten some of the + subqueries, but since const tables do not generate many duplicates, + it really doesn't matter that much whether they were pulled out or + not. + + All of this was done as fix for BUG#43768. + */ if (tbl->table->map & join->const_table_map) { pulled_tables |= tbl->table->map; DBUG_PRINT("info", ("Table %s pulled out (reason: constant)", tbl->table->alias)); } +#endif } } @@ -1048,6 +1069,7 @@ pointers. */ child_li.remove(); + sj_nest->nested_join->used_tables &= ~tbl->table->map; upper_join_list->push_back(tbl); tbl->join_list= upper_join_list; tbl->embedding= sj_nest->embedding; @@ -1104,20 +1126,20 @@ DBUG_ENTER("optimize_semijoin_nests"); List_iterator<TABLE_LIST> sj_list_it(join->select_lex->sj_nests); TABLE_LIST *sj_nest; - /* - The statement may have been executed with 'semijoin=on' earlier. - We need to verify that 'semijoin=on' still holds. - */ - if (optimizer_flag(join->thd, OPTIMIZER_SWITCH_SEMIJOIN) && - optimizer_flag(join->thd, OPTIMIZER_SWITCH_MATERIALIZATION)) + while ((sj_nest= sj_list_it++)) { - while ((sj_nest= sj_list_it++)) + /* semi-join nests with only constant tables are not valid */ + /// DBUG_ASSERT(sj_nest->sj_inner_tables & ~join->const_table_map); + + sj_nest->sj_mat_info= NULL; + /* + The statement may have been executed with 'semijoin=on' earlier. + We need to verify that 'semijoin=on' still holds. + */ + if (optimizer_flag(join->thd, OPTIMIZER_SWITCH_SEMIJOIN) && + optimizer_flag(join->thd, OPTIMIZER_SWITCH_MATERIALIZATION)) { - /* semi-join nests with only constant tables are not valid */ - DBUG_ASSERT(sj_nest->sj_inner_tables & ~join->const_table_map); - - sj_nest->sj_mat_info= NULL; - if (sj_nest->sj_inner_tables && /* not everything was pulled out */ + if ((sj_nest->sj_inner_tables & ~join->const_table_map) && /* not everything was pulled out */ !sj_nest->sj_subq_pred->is_correlated && sj_nest->sj_subq_pred->types_allow_materialization) { @@ -1128,7 +1150,7 @@ The best plan to run the subquery is now in join->best_positions, save it. */ - uint n_tables= my_count_bits(sj_nest->sj_inner_tables); + uint n_tables= my_count_bits(sj_nest->sj_inner_tables & ~join->const_table_map); SJ_MATERIALIZATION_INFO* sjm; if (!(sjm= new SJ_MATERIALIZATION_INFO) || !(sjm->positions= (POSITION*)join->thd->alloc(sizeof(POSITION)* @@ -1443,7 +1465,7 @@ new_join_tab->emb_sj_nest->nested_join->sj_corr_tables | new_join_tab->emb_sj_nest->nested_join->sj_depends_on; const table_map sj_inner_tables= - new_join_tab->emb_sj_nest->sj_inner_tables; + new_join_tab->emb_sj_nest->sj_inner_tables & ~join->const_table_map; /* Enter condition: === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-07 15:41:45 +0000 +++ b/sql/sql_select.cc 2010-03-14 17:54:12 +0000 @@ -5127,7 +5127,9 @@ /* number of tables that remain to be optimized */ n_tables= size_remain= my_count_bits(remaining_tables & (join->emb_sjm_nest? - join->emb_sjm_nest->sj_inner_tables : + (join->emb_sjm_nest->sj_inner_tables & + ~join->const_table_map) + : ~(table_map)0)); do { @@ -5387,7 +5389,7 @@ table_map allowed_tables= ~(table_map)0; if (join->emb_sjm_nest) - allowed_tables= join->emb_sjm_nest->sj_inner_tables; + allowed_tables= join->emb_sjm_nest->sj_inner_tables & ~join->const_table_map; for (JOIN_TAB **pos= join->best_ref + idx ; (s= *pos) ; pos++) {

1 0

[Maria-developers] Rev 2775: Apply fix by oystein.grovlen@sun.com 2010-03-12: in file:///home/psergey/dev/maria-5.3-subqueries-r7/
by Sergey Petrunya 13 Mar '10

13 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7/ ------------------------------------------------------------ revno: 2775 revision-id: psergey(a)askmonty.org-20100313211106-5xyfyl02gfenbi7f parent: psergey(a)askmonty.org-20100313200452-kq4dxayp7b45zum1 committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7 timestamp: Sun 2010-03-14 00:11:06 +0300 message: Apply fix by oystein.grovlen(a)sun.com 2010-03-12: Bug#48213 Materialized subselect crashes if using GEOMETRY type The problem occurred because during semi-join a materialized table was created which contained a GEOMETRY column, which is a specialized BLOB column. This caused an segmentation fault because such tables will have extra columns, and the semi-join code was not prepared for that. The solution is to disable materialization when Blob/Geometry columns would need to be materialized. Blob columns cannot be used for index look-up anyway, so it does not makes sense to use materialization. This fix implies that it is detected earlier that subquery materialization can not be used. The result of that is that in->exist optimization may be performed for such queries. Hence, extended query plans for such queries had to be updated. === modified file 'mysql-test/r/subselect_mat.result' --- a/mysql-test/r/subselect_mat.result 2010-03-13 20:04:52 +0000 +++ b/mysql-test/r/subselect_mat.result 2010-03-13 21:11:06 +0000 @@ -583,7 +583,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a1` in (select 1 AS `Not_used` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`)))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`)))) select left(a1,7), left(a2,7) from t1_16 where a1 in (select b1 from t2_16 where b1 > '0'); @@ -597,7 +597,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`) in (select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),<exists>(select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` where ((`test`.`t2_16`.`b1` > '0') and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) select left(a1,7), left(a2,7) from t1_16 where (a1,a2) in (select b1, b2 from t2_16 where b1 > '0'); @@ -625,7 +625,7 @@ 1 PRIMARY t1_16 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_16 ALL NULL NULL NULL NULL 3 100.00 Using filesort Warnings: -Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a1` in (select group_concat(`test`.`t2_16`.`b1` separator ',') AS `group_concat(b1)` from `test`.`t2_16` group by `test`.`t2_16`.`b2` having (<cache>(`test`.`t1_16`.`a1`) = <ref_null_helper>(group_concat(`test`.`t2_16`.`b1` separator ','))))) +Note 1003 select left(`test`.`t1_16`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_16`.`a2`,7) AS `left(a2,7)` from `test`.`t1_16` where <in_optimizer>(`test`.`t1_16`.`a1`,<exists>(select group_concat(`test`.`t2_16`.`b1` separator ',') AS `group_concat(b1)` from `test`.`t2_16` group by `test`.`t2_16`.`b2` having (<cache>(`test`.`t1_16`.`a1`) = <ref_null_helper>(group_concat(`test`.`t2_16`.`b1` separator ','))))) select left(a1,7), left(a2,7) from t1_16 where a1 in (select group_concat(b1) from t2_16 group by b2); @@ -662,7 +662,7 @@ 3 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 5 100.00 Using where; Using join buffer 4 SUBQUERY t3 ALL NULL NULL NULL NULL 4 100.00 Using where Warnings: -Note 1003 select `test`.`t1`.`a1` AS `a1`,`test`.`t1`.`a2` AS `a2` from `test`.`t1` where <in_optimizer>(concat(`test`.`t1`.`a1`,'x'),<exists>(select 1 AS `Not_used` from `test`.`t1_16` where (<in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),(`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`) in (select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` join `test`.`t2` where ((`test`.`t2`.`b2` = substr(`test`.`t2_16`.`b2`,1,6)) and <in_optimizer>(`test`.`t2`.`b1`,`test`.`t2`.`b1` in ( <materialize> (select `test`.`t3`.`c1` AS `c1` from `test`.`t3` where (`test`.`t3`.`c2` > '0') ), <primary_index_lookup>(`test`.`t2`.`b1` in <temporary table> on distinct_key where ((`test`.`t2`.`b1` = `materialized subselect`.`c1`))))) and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) and (<cache>(concat(`test`.`t1`.`a1`,'x')) = left(`test`.`t1_16`.`a1`,8))))) +Note 1003 select `test`.`t1`.`a1` AS `a1`,`test`.`t1`.`a2` AS `a2` from `test`.`t1` where <in_optimizer>(concat(`test`.`t1`.`a1`,'x'),<exists>(select 1 AS `Not_used` from `test`.`t1_16` where (<in_optimizer>((`test`.`t1_16`.`a1`,`test`.`t1_16`.`a2`),<exists>(select `test`.`t2_16`.`b1` AS `b1`,`test`.`t2_16`.`b2` AS `b2` from `test`.`t2_16` join `test`.`t2` where ((`test`.`t2`.`b2` = substr(`test`.`t2_16`.`b2`,1,6)) and <in_optimizer>(`test`.`t2`.`b1`,`test`.`t2`.`b1` in ( <materialize> (select `test`.`t3`.`c1` AS `c1` from `test`.`t3` where (`test`.`t3`.`c2` > '0') ), <primary_index_lookup>(`test`.`t2`.`b1` in <temporary table> on distinct_key where ((`test`.`t2`.`b1` = `materialized subselect`.`c1`))))) and (<cache>(`test`.`t1_16`.`a1`) = `test`.`t2_16`.`b1`) and (<cache>(`test`.`t1_16`.`a2`) = `test`.`t2_16`.`b2`)))) and (<cache>(concat(`test`.`t1`.`a1`,'x')) = left(`test`.`t1_16`.`a1`,8))))) drop table t1_16, t2_16, t3_16; set @blob_len = 512; set @suffix_len = @blob_len - @prefix_len; @@ -696,7 +696,7 @@ 1 PRIMARY t1_512 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_512 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>(`test`.`t1_512`.`a1`,`test`.`t1_512`.`a1` in (select 1 AS `Not_used` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`)))) +Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>(`test`.`t1_512`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`)))) select left(a1,7), left(a2,7) from t1_512 where a1 in (select b1 from t2_512 where b1 > '0'); @@ -710,7 +710,7 @@ 1 PRIMARY t1_512 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_512 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>((`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`),(`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`) in (select `test`.`t2_512`.`b1` AS `b1`,`test`.`t2_512`.`b2` AS `b2` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`) and (<cache>(`test`.`t1_512`.`a2`) = `test`.`t2_512`.`b2`)))) +Note 1003 select left(`test`.`t1_512`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_512`.`a2`,7) AS `left(a2,7)` from `test`.`t1_512` where <in_optimizer>((`test`.`t1_512`.`a1`,`test`.`t1_512`.`a2`),<exists>(select `test`.`t2_512`.`b1` AS `b1`,`test`.`t2_512`.`b2` AS `b2` from `test`.`t2_512` where ((`test`.`t2_512`.`b1` > '0') and (<cache>(`test`.`t1_512`.`a1`) = `test`.`t2_512`.`b1`) and (<cache>(`test`.`t1_512`.`a2`) = `test`.`t2_512`.`b2`)))) select left(a1,7), left(a2,7) from t1_512 where (a1,a2) in (select b1, b2 from t2_512 where b1 > '0'); @@ -789,7 +789,7 @@ 1 PRIMARY t1_1024 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1024 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>(`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a1` in (select 1 AS `Not_used` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`)))) +Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>(`test`.`t1_1024`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`)))) select left(a1,7), left(a2,7) from t1_1024 where a1 in (select b1 from t2_1024 where b1 > '0'); @@ -803,7 +803,7 @@ 1 PRIMARY t1_1024 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1024 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>((`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`),(`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`) in (select `test`.`t2_1024`.`b1` AS `b1`,`test`.`t2_1024`.`b2` AS `b2` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`) and (<cache>(`test`.`t1_1024`.`a2`) = `test`.`t2_1024`.`b2`)))) +Note 1003 select left(`test`.`t1_1024`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1024`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1024` where <in_optimizer>((`test`.`t1_1024`.`a1`,`test`.`t1_1024`.`a2`),<exists>(select `test`.`t2_1024`.`b1` AS `b1`,`test`.`t2_1024`.`b2` AS `b2` from `test`.`t2_1024` where ((`test`.`t2_1024`.`b1` > '0') and (<cache>(`test`.`t1_1024`.`a1`) = `test`.`t2_1024`.`b1`) and (<cache>(`test`.`t1_1024`.`a2`) = `test`.`t2_1024`.`b2`)))) select left(a1,7), left(a2,7) from t1_1024 where (a1,a2) in (select b1, b2 from t2_1024 where b1 > '0'); @@ -882,7 +882,7 @@ 1 PRIMARY t1_1025 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1025 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>(`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a1` in (select 1 AS `Not_used` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`)))) +Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>(`test`.`t1_1025`.`a1`,<exists>(select 1 AS `Not_used` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`)))) select left(a1,7), left(a2,7) from t1_1025 where a1 in (select b1 from t2_1025 where b1 > '0'); @@ -896,7 +896,7 @@ 1 PRIMARY t1_1025 ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2_1025 ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>((`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`),(`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`) in (select `test`.`t2_1025`.`b1` AS `b1`,`test`.`t2_1025`.`b2` AS `b2` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`) and (<cache>(`test`.`t1_1025`.`a2`) = `test`.`t2_1025`.`b2`)))) +Note 1003 select left(`test`.`t1_1025`.`a1`,7) AS `left(a1,7)`,left(`test`.`t1_1025`.`a2`,7) AS `left(a2,7)` from `test`.`t1_1025` where <in_optimizer>((`test`.`t1_1025`.`a1`,`test`.`t1_1025`.`a2`),<exists>(select `test`.`t2_1025`.`b1` AS `b1`,`test`.`t2_1025`.`b2` AS `b2` from `test`.`t2_1025` where ((`test`.`t2_1025`.`b1` > '0') and (<cache>(`test`.`t1_1025`.`a1`) = `test`.`t2_1025`.`b1`) and (<cache>(`test`.`t1_1025`.`a2`) = `test`.`t2_1025`.`b2`)))) select left(a1,7), left(a2,7) from t1_1025 where (a1,a2) in (select b1, b2 from t2_1025 where b1 > '0'); @@ -982,7 +982,7 @@ 1 PRIMARY t1bb ALL NULL NULL NULL NULL 3 100.00 Using where 2 DEPENDENT SUBQUERY t2bb ALL NULL NULL NULL NULL 3 100.00 Using where Warnings: -Note 1003 select conv(`test`.`t1bb`.`a1`,10,2) AS `bin(a1)`,`test`.`t1bb`.`a2` AS `a2` from `test`.`t1bb` where <in_optimizer>((`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`),(`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`) in (select `test`.`t2bb`.`b1` AS `b1`,`test`.`t2bb`.`b2` AS `b2` from `test`.`t2bb` where ((<cache>(`test`.`t1bb`.`a1`) = `test`.`t2bb`.`b1`) and (<cache>(`test`.`t1bb`.`a2`) = `test`.`t2bb`.`b2`)))) +Note 1003 select conv(`test`.`t1bb`.`a1`,10,2) AS `bin(a1)`,`test`.`t1bb`.`a2` AS `a2` from `test`.`t1bb` where <in_optimizer>((`test`.`t1bb`.`a1`,`test`.`t1bb`.`a2`),<exists>(select `test`.`t2bb`.`b1` AS `b1`,`test`.`t2bb`.`b2` AS `b2` from `test`.`t2bb` where ((<cache>(`test`.`t1bb`.`a1`) = `test`.`t2bb`.`b1`) and (<cache>(`test`.`t1bb`.`a2`) = `test`.`t2bb`.`b2`)))) select bin(a1), a2 from t1bb where (a1, a2) in (select b1, b2 from t2bb); === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-03-13 20:04:52 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-13 21:11:06 +0000 @@ -825,6 +825,127 @@ 2 drop table t1, t2, t3; # +# Bug#48213 Materialized subselect crashes if using GEOMETRY type +# +CREATE TABLE t1 ( +pk int, +a varchar(1), +b varchar(4), +c tinyblob, +d blob, +e mediumblob, +f longblob, +g tinytext, +h text, +i mediumtext, +j longtext, +k geometry, +PRIMARY KEY (pk) +); +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using MRR; Materialize +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`c` = `test`.`t1`.`c`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`d` = `test`.`t1`.`d`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`e` = `test`.`t1`.`e`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`f` = `test`.`t1`.`f`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`g` = `test`.`t1`.`g`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`h` = `test`.`t1`.`h`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`i` = `test`.`t1`.`i`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`j` = `test`.`t1`.`j`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1) +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`k` = `test`.`t1`.`k`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +pk +1 +2 +DROP TABLE t1, t2; +# End of Bug#48213 +# # Bug#49198 Wrong result for second call of procedure # with view in subselect. # === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-13 20:04:52 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-13 21:11:06 +0000 @@ -829,6 +829,127 @@ 2 drop table t1, t2, t3; # +# Bug#48213 Materialized subselect crashes if using GEOMETRY type +# +CREATE TABLE t1 ( +pk int, +a varchar(1), +b varchar(4), +c tinyblob, +d blob, +e mediumblob, +f longblob, +g tinytext, +h text, +i mediumtext, +j longtext, +k geometry, +PRIMARY KEY (pk) +); +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using MRR; Materialize +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`c` = `test`.`t1`.`c`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`d` = `test`.`t1`.`d`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +pk +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`e` = `test`.`t1`.`e`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`f` = `test`.`t1`.`f`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`g` = `test`.`t1`.`g`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`h` = `test`.`t1`.`h`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`i` = `test`.`t1`.`i`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`j` = `test`.`t1`.`j`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +pk +1 +2 +EXPLAIN EXTENDED SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t1 ALL NULL NULL NULL NULL 2 100.00 +1 PRIMARY t2 range PRIMARY PRIMARY 4 NULL 2 100.00 Using index condition; Using where; Using MRR; FirstMatch(t1); Using join buffer +Warnings: +Note 1003 select `test`.`t1`.`pk` AS `pk` from `test`.`t1` semi join (`test`.`t2`) where ((`test`.`t2`.`k` = `test`.`t1`.`k`) and (`test`.`t2`.`b` = `test`.`t1`.`b`) and (`test`.`t2`.`pk` > 0)) +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +pk +1 +2 +DROP TABLE t1, t2; +# End of Bug#48213 +# # Bug#49198 Wrong result for second call of procedure # with view in subselect. # === modified file 'mysql-test/t/subselect_sj.test' --- a/mysql-test/t/subselect_sj.test 2010-03-13 20:04:52 +0000 +++ b/mysql-test/t/subselect_sj.test 2010-03-13 21:11:06 +0000 @@ -729,6 +729,86 @@ drop table t1, t2, t3; --echo # +--echo # Bug#48213 Materialized subselect crashes if using GEOMETRY type +--echo # + +CREATE TABLE t1 ( + pk int, + a varchar(1), + b varchar(4), + c tinyblob, + d blob, + e mediumblob, + f longblob, + g tinytext, + h text, + i mediumtext, + j longtext, + k geometry, + PRIMARY KEY (pk) +); + +INSERT INTO t1 VALUES (1,'o','ffff','ffff','ffoo','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff', 'ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); + +CREATE TABLE t2 LIKE t1; +INSERT INTO t2 VALUES (1,'i','iiii','iiii','iiii','iiii','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')), (2,'f','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff','ffff',GeomFromText('POLYGON((0 0, 0 2, 2 2, 2 0, 0 0))')); + +# Test that materialization is skipped for semijoins where materialized +# table would contain GEOMETRY or different kinds of BLOB/TEXT columns +let $query= +SELECT pk FROM t1 WHERE (a, b) IN (SELECT a, b FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, c) IN (SELECT b, c FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, d) IN (SELECT b, d FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, e) IN (SELECT b, e FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, f) IN (SELECT b, f FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, g) IN (SELECT b, g FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, h) IN (SELECT b, h FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, i) IN (SELECT b, i FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, j) IN (SELECT b, j FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +let $query= +SELECT pk FROM t1 WHERE (b, k) IN (SELECT b, k FROM t2 WHERE pk > 0); +eval EXPLAIN EXTENDED $query; +eval $query; + +DROP TABLE t1, t2; +--echo # End of Bug#48213 + +--echo # --echo # Bug#49198 Wrong result for second call of procedure --echo # with view in subselect. --echo # === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-13 20:04:52 +0000 +++ b/sql/opt_subselect.cc 2010-03-13 21:11:06 +0000 @@ -322,7 +322,13 @@ default: ;/* suitable for materialization */ } + + // Materialization does not work with BLOB columns + if (inner->field_type() == MYSQL_TYPE_BLOB || + inner->field_type() == MYSQL_TYPE_GEOMETRY) + DBUG_RETURN(FALSE); } + in_subs->types_allow_materialization= TRUE; in_subs->sjm_scan_allowed= all_are_fields; DBUG_PRINT("info",("subquery_types_allow_materialization: ok, allowed"));

1 0

Re: [Maria-developers] bzr commit into mysql-5.4 branch (epotemkin:2814) Bug#45174
by Sergey Petrunya 13 Mar '10

13 Mar '10

Hi! I can't offer any testcases but I think this patch has several issues. See below. On Tue, Oct 13, 2009 at 09:38:52AM +0000, Evgeny Potemkin wrote: > #At file:///work/bzrroot/45174-bug-azalea/ based on revid:alik@sun.com-20090702085822-8svd0aslr7qnddbb > > 2814 Evgeny Potemkin 2009-10-13 > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > > When a subquery is a subject to a semi-join optimization its tables are > merged to the upper query and later they treated as usual tables. > This allows a bunch of optimizations to be applied, equality > propagation is among them. Equality propagation is done after query execution > plan is chosen. It substitutes fields from tables being retrieved later for > fields from tables being retrieved earlier. However it can't be applied as is > to any semi-join table. > The semi-join materialization strategy differs from other semi-join > strategies that the data from materialized semi-join tables isn't used > directly but saved to a temporary table first. The materialization isn't > isolated is a separate step, it is done inline within the nested loop execution. > When it comes to fetch rows from the first table in > the block of materialized semi-join tables they are isolated and the > sub_select function is called to materialize result and save it in the > semi-join result table. Materialization is done once and later data from the > semi-join result table is used. > Due to this we can't substitute fields that belong to the semi-join > for fields from outer query and vice versa. > > Example: suppose we have a join order: > > ot1 ot2 SJ-Mat(it1 it2 it3) ot3 > > and equality ot2.col = it1.col = it2.col > If we're looking for best substitute for 'it2.col', we should pick it1.col > and not ot2.col. > > For a field that is not in a materialized semi-join we must pick a field > that's not embedded in a materialized semi-join. > > Example: suppose we have a join order: > > SJ-Mat(it1 it2) ot1 ot2 > > and equality ot2.col = ot1.col = it2.col > If we're looking for best substitute for 'ot2.col', we should pick ot1.col > and not it2.col, because when we run a join between ot1 and ot2 > execution of SJ-Mat(...) has already finished and we can't rely on the value > of it*.*. > > Now the Item_equal::get_first function accepts as a parameter a field being > substituted and checks whether it belongs to a materialized semi-join. > Depending on the check result a field to substitute for or NULL is returned. > > The sj_strategy field is added to the st_join_table structure. It's a copy of the > POSITION::sj_strategy field and is used to easy checks. > @ mysql-test/r/subselect_sj.result > A test case added for the bug#45174. > @ mysql-test/r/subselect_sj_jcl6.result > A test case added for the bug#45174. > @ mysql-test/t/subselect_sj.test > A test case added for the bug#45174. > @ sql/item.cc > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > Now the Item_equal::get_first function accepts as a parameter a field being > substituted. > @ sql/item_cmpfunc.cc > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > > Now the Item_equal::get_first function accepts a field being substituted and > checks whether it belongs to a materialized semi-join. Depending on the check > result a field to substitute for or NULL is returned. > @ sql/item_cmpfunc.h > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > > Now the Item_equal::get_first function accepts as a parameter a field being > substituted. > @ sql/sql_select.cc > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > The is_sj_materialization_strategy method is added to the JOIN_TAB class to > check whether JOIN_TAB belongs to a materialized semi-join. > @ sql/sql_select.h > Bug#45174: Incorrectly applied equality propagation caused wrong result > on a query with a materialized semi-join. > > The sj_strategy field is added to the st_join_table structure. It's a copy of the > POSITION::sj_strategy field and is used to easy checks. > > modified: > mysql-test/r/subselect_sj.result > mysql-test/r/subselect_sj_jcl6.result > mysql-test/t/subselect_sj.test > sql/item.cc > sql/item_cmpfunc.cc > sql/item_cmpfunc.h > sql/sql_select.cc > sql/sql_select.h > === modified file 'sql/item.cc' > --- a/sql/item.cc 2009-06-09 16:53:34 +0000 > +++ b/sql/item.cc 2009-10-13 09:38:46 +0000 > @@ -4883,7 +4883,7 @@ Item *Item_field::replace_equal_field(uc > return this; > return const_item; > } > - Item_field *subst= item_equal->get_first(); > + Item_field *subst= item_equal->get_first(this); > if (subst && field->table != subst->field->table && !field->eq(subst->field)) > return subst; > } > > === modified file 'sql/item_cmpfunc.cc' > --- a/sql/item_cmpfunc.cc 2009-06-09 16:53:34 +0000 > +++ b/sql/item_cmpfunc.cc 2009-10-13 09:38:46 +0000 > @@ -5376,7 +5376,7 @@ longlong Item_equal::val_int() > > void Item_equal::fix_length_and_dec() > { > - Item *item= get_first(); > + Item *item= get_first(NULL); > eval_item= cmp_item::get_comparator(item->result_type(), > item->collation.collation); > } > @@ -5439,3 +5439,115 @@ void Item_equal::print(String *str, enum > str->append(')'); > } > > + > +/* > + @brief Get the first equal field of multiple equality. > + @param[in] field the field to get equal field to > + > + @details Get the first field of multiple equality that is equal to the > + given field. In order to make semi-join materialization strategy work > + correctly we can't propagate equal fields from upper select to a > + materialized semi-join. > + Thus the fields is returned according to following rules: > + > + 1) If the given field belongs to a semi-join then the first field in > + multiple equality which belong to the same semi-join is returned. > + Otherwise NULL is returned. > + 2) If the given field doesn't belong to a semi-join then > + the first field in the multiple equality that doesn't belong to any > + semi-join is returned. > + If all fields in the equality are belong to semi-join(s) then NULL > + is returned. > + 3) If no field is given then the first field in the multiple equality > + is returned without regarding whether it belongs to a semi-join or not. > + > + @retval Found first field in the multiple equality. > + @retval 0 if no field found. > +*/ > + > +Item_field* Item_equal::get_first(Item_field *field) > +{ > + List_iterator<Item_field> it(fields); > + Item_field *item; > + JOIN_TAB *field_tab; > + > + if (!field) > + return fields.head(); > + /* > + Of all equal fields, return the first one we can use. Normally, this is the > + field which belongs to the table that is the first in the join order. > + > + There is one exception to this: When semi-join materialization strategy is > + used, and the given field belongs to a table within the semi-join nest, we > + must pick the first field in the semi-join nest. > + > + Example: suppose we have a join order: > + > + ot1 ot2 SJ-Mat(it1 it2 it3) ot3 > + > + and equality ot2.col = it1.col = it2.col > + If we're looking for best substitute for 'it2.col', we should pick it1.col > + and not ot2.col. > + */ > + > + field_tab= field->field->table->reginfo.join_tab; > + if (field_tab->sj_strategy == SJ_OPT_MATERIALIZE || > + field_tab->sj_strategy == SJ_OPT_MATERIALIZE_SCAN) > + { > + /* > + It's a field from an materialized semi-join. We can substitute it only > + for a field from the same semi-join. > + */ > + JOIN_TAB *first; > + JOIN *join= field_tab->join; > + uint tab_idx= field_tab - field_tab->join->join_tab; > + /* Find first table of this semi-join. */ > + for (int i=tab_idx; i >= join->const_tables; i--) > + { > + if (join->best_positions[i].sj_strategy == SJ_OPT_MATERIALIZE || > + join->best_positions[i].sj_strategy == SJ_OPT_MATERIALIZE_SCAN) > + first= join->join_tab + i; > + else > + // Found first tab that doesn't belong to current SJ. > + break; > + } > + /* Find an item to substitute for. */ > + while ((item= it++)) > + { > + if (item->field->table->reginfo.join_tab >= first) > + { > + /* > + If we found given field then return NULL to avoid unnecessary > + substitution. > + */ > + return (item != field) ? item : NULL; > + } > + } > + } > + else > + { > + /* > + The field is not in SJ-Materialization nest. We must return the first > + field that's not embedded in a SJ-Materialization nest. > + Example: suppose we have a join order: > + > + SJ-Mat(it1 it2) ot1 ot2 > + > + and equality ot2.col = ot1.col = it2.col > + If we're looking for best substitute for 'ot2.col', we should pick ot1.col > + and not it2.col, because when we run a join between ot1 and ot2 > + execution of SJ-Mat(...) has already finished and we can't rely on the > + value of it*.*. This can cause cross-join to be computed between materialization result and table it1. Actually, substitution with table it2 should be fine, as SJ-Materialization-Scan (and this example cannot be lookup) will 'unpack' column value to it2.col when doing the scan of the materialized temptable. I've wrote up my understanding of the problem here (with pics, so on the wiki): http://askmonty.org/wiki/EqualityPropagationAndEqualityPropagationAndSemiJo… > + */ > + while ((item= it++)) > + { > + field_tab= item->field->table->reginfo.join_tab; > + if (!(field_tab->sj_strategy == SJ_OPT_MATERIALIZE || > + field_tab->sj_strategy == SJ_OPT_MATERIALIZE_SCAN)) This is a wrong way to check if a field is inside SJ-Materialization nest. The condition is true only for the first table in SJ-Materialization nest, while we need to catch *any* SJ-Mat-inner table. the correct way to check this is as follows: field_tab->pos_in_table_list->embedding && field_tab->pos_in_table_list->embedding->sj_mat && field_tab->pos_in_table_list->embedding->sj_mat->is_used > + return item; > + } > + } > + // Shouldn't get here. > + DBUG_ASSERT(0); > + return NULL; > +} > > === modified file 'sql/item_cmpfunc.h' > --- a/sql/item_cmpfunc.h 2009-01-26 16:03:39 +0000 > +++ b/sql/item_cmpfunc.h 2009-10-13 09:38:46 +0000 > @@ -1592,7 +1592,7 @@ public: > void add(Item_field *f); > uint members(); > bool contains(Field *field); > - Item_field* get_first() { return fields.head(); } > + Item_field* get_first(Item_field *field); > void merge(Item_equal *item); > void update_const(); > enum Functype functype() const { return MULT_EQUAL_FUNC; } > > === modified file 'sql/sql_select.cc' > --- a/sql/sql_select.cc 2009-06-30 08:03:05 +0000 > +++ b/sql/sql_select.cc 2009-10-13 09:38:46 +0000 > @@ -7911,6 +7911,7 @@ static void fix_semijoin_strategies_for_ > if (tablenr != first) > pos->sj_strategy= SJ_OPT_NONE; > remaining_tables |= s->table->map; > + s->sj_strategy= pos->sj_strategy; > } > } > > @@ -11706,7 +11707,7 @@ Item *eliminate_item_equal(COND *cond, C > head= item_const; > else > { > - head= item_equal->get_first(); > + head= item_equal->get_first(NULL); > it++; > } > Item_field *item_field; > > === modified file 'sql/sql_select.h' > --- a/sql/sql_select.h 2009-05-07 20:48:24 +0000 > +++ b/sql/sql_select.h 2009-10-13 09:38:46 +0000 > @@ -274,6 +274,13 @@ typedef struct st_join_table > /* NestedOuterJoins: Bitmap of nested joins this table is part of */ > nested_join_map embedding_map; > > + /* > + Semi-join strategy to be used for this join table. This is a copy of > + POSITION::sj_strategy field. This field is set up by the > + fix_semijion_strategies_for_picked_join_order. > + */ > + uint sj_strategy; > + > void cleanup(); > inline bool is_using_loose_index_scan() > { > BR Sergey -- Sergey Petrunia, Software Developer Monty Program AB, http://askmonty.org Blog: http://s.petrunia.net/blog

1 0

[Maria-developers] Rev 2774: BUG#45174: XOR in subqueries produces differing results in 5.1 and 5.4 in file:///home/psergey/dev/maria-5.3-subqueries-r7/
by Sergey Petrunya 13 Mar '10

13 Mar '10

At file:///home/psergey/dev/maria-5.3-subqueries-r7/ ------------------------------------------------------------ revno: 2774 revision-id: psergey(a)askmonty.org-20100313200452-kq4dxayp7b45zum1 parent: psergey(a)askmonty.org-20100307154145-ksby2b1l0sqm1xne committer: Sergey Petrunya <psergey(a)askmonty.org> branch nick: maria-5.3-subqueries-r7 timestamp: Sat 2010-03-13 23:04:52 +0300 message: BUG#45174: XOR in subqueries produces differing results in 5.1 and 5.4 BUG#50019: Wrong result for IN-subquery with materialization - Fix equality substitution in presense of semi-join materialization, lookup and scan variants (started off from fix by Evgen Potemkin, then modified it to work in all cases) === modified file 'mysql-test/r/subselect_mat.result' --- a/mysql-test/r/subselect_mat.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/subselect_mat.result 2010-03-13 20:04:52 +0000 @@ -1219,3 +1219,28 @@ pk 2 DROP TABLE t1, t2; +# +# BUG#50019: Wrong result for IN-subquery with materialization +# +create table t1(i int); +insert into t1 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t2(i int); +insert into t2 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t3(i int); +insert into t3 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +i +1 +2 +3 +4 +set @save_optimizer_switch=@@optimizer_switch; +set session optimizer_switch='materialization=off'; +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +i +1 +2 +3 +4 +set session optimizer_switch=@save_optimizer_switch; +drop table t1, t2, t3; === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-02-24 11:33:42 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-13 20:04:52 +0000 @@ -871,3 +871,39 @@ DROP TABLE t1, t2, t3; DROP VIEW v2, v3; # End of Bug#49198 +# +# Bug#45174: Incorrectly applied equality propagation caused wrong +# result on a query with a materialized semi-join. +# +CREATE TABLE `t1` ( +`pk` int(11) NOT NULL AUTO_INCREMENT, +`varchar_key` varchar(1) NOT NULL, +`varchar_nokey` varchar(1) NOT NULL, +PRIMARY KEY (`pk`), +KEY `varchar_key` (`varchar_key`) +); +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); +CREATE TABLE `t2` ( +`varchar_nokey` varchar(1) NOT NULL +); +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t2 ALL NULL NULL NULL NULL 18 100.00 +1 PRIMARY t1 ALL varchar_key NULL NULL NULL 15 100.00 Using where; Materialize +Warnings: +Note 1003 select `test`.`t2`.`varchar_nokey` AS `varchar_nokey` from `test`.`t2` semi join (`test`.`t1`) where ((`test`.`t1`.`varchar_nokey` = `test`.`t1`.`varchar_key`) and ((`test`.`t1`.`varchar_nokey` < 'n') xor `test`.`t1`.`pk`)) +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +varchar_nokey +DROP TABLE t1, t2; +# End of the test for bug#45174. === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-07 15:41:45 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-13 20:04:52 +0000 @@ -876,6 +876,42 @@ DROP VIEW v2, v3; # End of Bug#49198 # +# Bug#45174: Incorrectly applied equality propagation caused wrong +# result on a query with a materialized semi-join. +# +CREATE TABLE `t1` ( +`pk` int(11) NOT NULL AUTO_INCREMENT, +`varchar_key` varchar(1) NOT NULL, +`varchar_nokey` varchar(1) NOT NULL, +PRIMARY KEY (`pk`), +KEY `varchar_key` (`varchar_key`) +); +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); +CREATE TABLE `t2` ( +`varchar_nokey` varchar(1) NOT NULL +); +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +id select_type table type possible_keys key key_len ref rows filtered Extra +1 PRIMARY t2 ALL NULL NULL NULL NULL 18 100.00 +1 PRIMARY t1 ALL varchar_key NULL NULL NULL 15 100.00 Using where; Materialize +Warnings: +Note 1003 select `test`.`t2`.`varchar_nokey` AS `varchar_nokey` from `test`.`t2` semi join (`test`.`t1`) where ((`test`.`t1`.`varchar_nokey` = `test`.`t1`.`varchar_key`) and ((`test`.`t1`.`varchar_nokey` < 'n') xor `test`.`t1`.`pk`)) +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; +varchar_nokey +DROP TABLE t1, t2; +# End of the test for bug#45174. +# # BUG#49129: Wrong result with IN-subquery with join_cache_level=6 and firstmatch=off # CREATE TABLE t0 (a INT); === modified file 'mysql-test/t/subselect_mat.test' --- a/mysql-test/t/subselect_mat.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect_mat.test 2010-03-13 20:04:52 +0000 @@ -889,3 +889,19 @@ SELECT pk FROM t1 WHERE (b,c,d) IN (SELECT b,c,d FROM t2 WHERE pk > 0); DROP TABLE t1, t2; +--echo # +--echo # BUG#50019: Wrong result for IN-subquery with materialization +--echo # +create table t1(i int); +insert into t1 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t2(i int); +insert into t2 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +create table t3(i int); +insert into t3 values (1), (2), (3), (4), (5), (6), (7), (8), (9), (10); +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +set @save_optimizer_switch=@@optimizer_switch; +set session optimizer_switch='materialization=off'; +select * from t1 where t1.i in (select t2.i from t2 join t3 where t2.i + t3.i = 5); +set session optimizer_switch=@save_optimizer_switch; +drop table t1, t2, t3; + === modified file 'mysql-test/t/subselect_sj.test' --- a/mysql-test/t/subselect_sj.test 2010-02-24 11:33:42 +0000 +++ b/mysql-test/t/subselect_sj.test 2010-03-13 20:04:52 +0000 @@ -770,3 +770,42 @@ DROP VIEW v2, v3; --echo # End of Bug#49198 + +--echo # +--echo # Bug#45174: Incorrectly applied equality propagation caused wrong +--echo # result on a query with a materialized semi-join. +--echo # + +CREATE TABLE `t1` ( + `pk` int(11) NOT NULL AUTO_INCREMENT, + `varchar_key` varchar(1) NOT NULL, + `varchar_nokey` varchar(1) NOT NULL, + PRIMARY KEY (`pk`), + KEY `varchar_key` (`varchar_key`) +); + +INSERT INTO `t1` VALUES (11,'m','m'),(12,'j','j'),(13,'z','z'),(14,'a','a'),(15,'',''),(16,'e','e'),(17,'t','t'),(19,'b','b'),(20,'w','w'),(21,'m','m'),(23,'',''),(24,'w','w'),(26,'e','e'),(27,'e','e'),(28,'p','p'); + +CREATE TABLE `t2` ( + `varchar_nokey` varchar(1) NOT NULL +); + +INSERT INTO `t2` VALUES ('v'),('u'),('n'),('l'),('h'),('u'),('n'),('j'),('k'),('e'),('i'),('u'),('n'),('b'),('x'),(''),('q'),('u'); + +EXPLAIN EXTENDED SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; + +SELECT varchar_nokey +FROM t2 +WHERE ( `varchar_nokey` , `varchar_nokey` ) IN ( +SELECT `varchar_key` , `varchar_nokey` +FROM t1 +WHERE `varchar_nokey` < 'n' XOR `pk` ) ; + +DROP TABLE t1, t2; + +--echo # End of the test for bug#45174. === modified file 'sql/item.cc' --- a/sql/item.cc 2010-02-24 11:33:42 +0000 +++ b/sql/item.cc 2010-03-13 20:04:52 +0000 @@ -4761,7 +4761,7 @@ return this; return const_item; } - Item_field *subst= item_equal->get_first(); + Item_field *subst= item_equal->get_first(this); if (subst && field->table != subst->field->table && !field->eq(subst->field)) return subst; } === modified file 'sql/item_cmpfunc.cc' --- a/sql/item_cmpfunc.cc 2010-02-17 10:05:27 +0000 +++ b/sql/item_cmpfunc.cc 2010-03-13 20:04:52 +0000 @@ -5369,7 +5369,7 @@ void Item_equal::fix_length_and_dec() { - Item *item= get_first(); + Item *item= get_first(NULL); eval_item= cmp_item::get_comparator(item->result_type(), item->collation.collation); } @@ -5432,3 +5432,128 @@ str->append(')'); } + +/* + @brief Get the first equal field of multiple equality. + @param[in] field the field to get equal field to + + @details Get the first field of multiple equality that is equal to the + given field. In order to make semi-join materialization strategy work + correctly we can't propagate equal fields from upper select to a + materialized semi-join. + Thus the fields is returned according to following rules: + + 1) If the given field belongs to a semi-join then the first field in + multiple equality which belong to the same semi-join is returned. + Otherwise NULL is returned. + 2) If the given field doesn't belong to a semi-join then + the first field in the multiple equality that doesn't belong to any + semi-join is returned. + If all fields in the equality are belong to semi-join(s) then NULL + is returned. + 3) If no field is given then the first field in the multiple equality + is returned without regarding whether it belongs to a semi-join or not. + + @retval Found first field in the multiple equality. + @retval 0 if no field found. +*/ + +Item_field* Item_equal::get_first(Item_field *field) +{ + List_iterator<Item_field> it(fields); + Item_field *item; + JOIN_TAB *field_tab; + + if (!field) + return fields.head(); + + /* + Of all equal fields, return the first one we can use. Normally, this is the + field which belongs to the table that is the first in the join order. + + There is one exception to this: When semi-join materialization strategy is + used, and the given field belongs to a table within the semi-join nest, we + must pick the first field in the semi-join nest. + + Example: suppose we have a join order: + + ot1 ot2 SJ-Mat(it1 it2 it3) ot3 + + and equality ot2.col = it1.col = it2.col + If we're looking for best substitute for 'it2.col', we should pick it1.col + and not ot2.col. + + eliminate_item_equal() also has code that deals with equality substitution + in presense of SJM nests. + */ + + field_tab= field->field->table->reginfo.join_tab; + + TABLE_LIST *emb_nest= field->field->table->pos_in_table_list->embedding; + + if (emb_nest && emb_nest->sj_mat_info && emb_nest->sj_mat_info->is_used) + { + /* + It's a field from an materialized semi-join. We can substitute it only + for a field from the same semi-join. + */ + JOIN_TAB *first; + JOIN *join= field_tab->join; + uint tab_idx= field_tab - field_tab->join->join_tab; + + /* Find the first table of this semi-join nest */ + for (uint i= tab_idx; i != join->const_tables; i--) + { + if (join->join_tab[i].table->map & emb_nest->sj_inner_tables) + first= join->join_tab + i; + else + // Found first tab that doesn't belong to current SJ. + break; + } + /* Find an item to substitute for. */ + while ((item= it++)) + { + if (item->field->table->reginfo.join_tab >= first) + { + /* + If we found given field then return NULL to avoid unnecessary + substitution. + */ + return (item != field) ? item : NULL; + } + } + } + else + { +#if 0 + /* + The field is not in SJ-Materialization nest. We must return the first + field that's not embedded in a SJ-Materialization nest. + Example: suppose we have a join order: + + SJ-Mat(it1 it2) ot1 ot2 + + and equality ot2.col = ot1.col = it2.col + If we're looking for best substitute for 'ot2.col', we should pick ot1.col + and not it2.col, because when we run a join between ot1 and ot2 + execution of SJ-Mat(...) has already finished and we can't rely on the + value of it*.*. + psergey-fix-fix: ^^ THAT IS INCORRECT ^^. Pick the first, whatever that + is. + */ + while ((item= it++)) + { + TABLE_LIST *emb_nest= item->field->table->pos_in_table_list->embedding; + if (!emb_nest || !emb_nest->sj_mat_info || + !emb_nest->sj_mat_info->is_used) + { + return item; + } + } +#endif + return fields.head(); + } + // Shouldn't get here. + DBUG_ASSERT(0); + return NULL; +} === modified file 'sql/item_cmpfunc.h' --- a/sql/item_cmpfunc.h 2010-02-17 10:05:27 +0000 +++ b/sql/item_cmpfunc.h 2010-03-13 20:04:52 +0000 @@ -1589,7 +1589,7 @@ void add(Item_field *f); uint members(); bool contains(Field *field); - Item_field* get_first() { return fields.head(); } + Item_field* get_first(Item_field *field); uint n_fields() { return fields.elements; } void merge(Item_equal *item); void update_const(); === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-07 15:41:45 +0000 +++ b/sql/opt_subselect.cc 2010-03-13 20:04:52 +0000 @@ -2159,6 +2159,8 @@ if (tablenr != first) pos->sj_strategy= SJ_OPT_NONE; remaining_tables |= s->table->map; + //s->sj_strategy= pos->sj_strategy; + join->join_tab[first].sj_strategy= join->best_positions[first].sj_strategy; } } === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2010-03-07 15:41:45 +0000 +++ b/sql/sql_select.cc 2010-03-13 20:04:52 +0000 @@ -8867,6 +8867,15 @@ } +static TABLE_LIST* embedding_sjm(Item_field *item_field) +{ + TABLE_LIST *nest= item_field->field->table->pos_in_table_list->embedding; + if (nest && nest->sj_mat_info && nest->sj_mat_info->is_used) + return nest; + else + return NULL; +} + /** Generate minimal set of simple equalities equivalent to a multiple equality. @@ -8900,6 +8909,23 @@ So only t1.a=t3.c should be left in the lower level. If cond is equal to 0, then not more then one equality is generated and a pointer to it is returned as the result of the function. + + Equality substutution and semi-join materialization nests: + + In case join order looks like this: + + outer_tbl1 outer_tbl2 SJM (inner_tbl1 inner_tbl2) outer_tbl3 + + We must not construct equalities like + + outer_tbl1.col = inner_tbl1.col + + because they would get attached to inner_tbl1 and will get evaluated + during materialization phase, when we don't have current value of + outer_tbl1.col. + + Item_equal::get_first() also takes similar measures for dealing with + equality substitution in presense of SJM nests. @return - The condition with generated simple equalities or @@ -8917,18 +8943,44 @@ Item *item_const= item_equal->get_const(); Item_equal_iterator it(*item_equal); Item *head; + TABLE_LIST *current_sjm= NULL; + Item *current_sjm_head= NULL; + + /* + Pick the "head" item: the constant one or the first in the join order + that's not inside some SJM nest. + */ if (item_const) head= item_const; else { - head= item_equal->get_first(); + TABLE_LIST *emb_nest; + Item_field *item_field; + head= item_field= item_equal->get_first(NULL); it++; + if ((emb_nest= embedding_sjm(item_field))) + { + current_sjm= emb_nest; + current_sjm_head= head; + } } + Item_field *item_field; + /* + For each other item, generate "item=head" equality (except the tables that + are within SJ-Materialization nests, for those "head" is defined + differently) + */ while ((item_field= it++)) { Item_equal *upper= item_field->find_item_equal(upper_levels); Item_field *item= item_field; + TABLE_LIST *field_sjm= embedding_sjm(item_field); + + /* + Check if "item_field=head" equality is already guaranteed to be true + on upper AND-levels. + */ if (upper) { if (item_const && upper->get_const()) @@ -8943,65 +8995,29 @@ } } } - if (item == item_field) + + bool produce_equality= test(item == item_field); + if (!item_const && field_sjm && field_sjm != current_sjm) + { + /* Entering an SJM nest */ + current_sjm_head= item_field; + if (!field_sjm->sj_mat_info->is_sj_scan) + produce_equality= FALSE; + } + + if (produce_equality) { if (eq_item) eq_list.push_back(eq_item); - /* - item_field might refer to a table that is within a semi-join - materialization nest. In that case, the join order looks like this: - - outer_tbl1 outer_tbl2 SJM (inner_tbl1 inner_tbl2) outer_tbl3 - - We must not construct equalities like - - outer_tbl1.col = inner_tbl1.col - - because they would get attached to inner_tbl1 and will get evaluated - during materialization phase, when we don't have current value of - outer_tbl1.col. - */ - TABLE_LIST *emb_nest= - item_field->field->table->pos_in_table_list->embedding; - if (!item_const && emb_nest && emb_nest->sj_mat_info && - emb_nest->sj_mat_info->is_used) - { - /* - Find the first equal expression that refers to a table that is - within the semijoin nest. If we can't find it, do nothing - */ - List_iterator<Item_field> fit(item_equal->fields); - Item_field *head_in_sjm; - bool found= FALSE; - while ((head_in_sjm= fit++)) - { - if (head_in_sjm->used_tables() & emb_nest->sj_inner_tables) - { - if (head_in_sjm == item_field) - { - /* This is the first table inside the semi-join*/ - eq_item= new Item_func_eq(item_field, head); - /* Tell make_cond_for_table don't use this. */ - eq_item->marker=3; - } - else - { - eq_item= new Item_func_eq(item_field, head_in_sjm); - found= TRUE; - } - break; - } - } - if (!found) - continue; - } - else - eq_item= new Item_func_eq(item_field, head); + + eq_item= new Item_func_eq(item_field, current_sjm? current_sjm_head: head); + if (!eq_item) return 0; eq_item->set_cmp_func(); eq_item->quick_fix_field(); } + current_sjm= field_sjm; } if (!cond && !eq_list.head()) === modified file 'sql/sql_select.h' --- a/sql/sql_select.h 2010-03-05 18:54:48 +0000 +++ b/sql/sql_select.h 2010-03-13 20:04:52 +0000 @@ -279,6 +279,13 @@ /* NestedOuterJoins: Bitmap of nested joins this table is part of */ nested_join_map embedding_map; + /* + Semi-join strategy to be used for this join table. This is a copy of + POSITION::sj_strategy field. This field is set up by the + fix_semijion_strategies_for_picked_join_order. + */ + uint sj_strategy; + void cleanup(); inline bool is_using_loose_index_scan() {

1 0

Re: [Maria-developers] Need help with packaging for MariaDB 5.2
by Kristian Nielsen 13 Mar '10

13 Mar '10

"Adam M. Dutko" <dutko.adam(a)gmail.com> writes: > I've packaged RPMs before if you'd like me to take a stab at it. Do you > have an existing spec file? Any help would be highly appreciated, thanks! I guess we just need to coordinate to not duplicate efforts. The spec file is in this repository on Launchpad: lp:~ourdelta-core/ourdelta/ourdelta-mariadb-5.2 The file in that repository is bakery/mysql51-ourdelta-centos.spec Maybe we should make a separate copy of that for 5.2, not sure. I guess at this point the main issues is to handle any dependency headers correctly, provides:, replaces:, depends: correctly. I'm pretty blank in knowledge about that area for .rpm. I'll start looking at the .deb stuff following Arjen's suggestions. - Kristian.

1 0

[Maria-developers] Updated (by Knielsen): Update packaging scripts for MariaDB 5.2 (88)
by worklog-noreply＠askmonty.org 13 Mar '10

13 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Update packaging scripts for MariaDB 5.2 CREATION DATE..: Sat, 27 Feb 2010, 16:39 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 88 (http://askmonty.org/worklog/?tid=88) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 30 (hours remain) ORIG. ESTIMATE.: 30 PROGRESS NOTES: -=-=(Knielsen - Sat, 13 Mar 2010, 08:14)=-=- Low Level Design modified. --- /tmp/wklog.88.old.22266 2010-03-13 08:14:47.000000000 +0000 +++ /tmp/wklog.88.new.22266 2010-03-13 08:14:47.000000000 +0000 @@ -1 +1,11 @@ +Some of the tasks that need to be done. + + - Setup a 5.2 version of .deb files and .rpm spec file. + + - Rename 5.1->5.2 in relevant places. + + - Fix provides: / replaces: and similar to ensure proper upgrade from mysql + 5.0/5.1 and mariadb 5.1. + + - Setup Buildbot upgrade test from MariaDB 5.1.42 -=-=(Guest - Sat, 13 Mar 2010, 08:12)=-=- Category updated. --- /tmp/wklog.88.old.22167 2010-03-13 08:12:01.000000000 +0000 +++ /tmp/wklog.88.new.22167 2010-03-13 08:12:01.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint DESCRIPTION: The packaging scripts need to be updated to work for MariaDB 5.2 Currently, 5.2 package builds fail in Buildbot. The .debs are missing a debian-5.2 subdirectory. The .rpm also need to be checked. Buildbot needs to be updated to do the new upgrade tests (mariadb-5.1 -> mariadb 5.2) LOW-LEVEL DESIGN: Some of the tasks that need to be done. - Setup a 5.2 version of .deb files and .rpm spec file. - Rename 5.1->5.2 in relevant places. - Fix provides: / replaces: and similar to ensure proper upgrade from mysql 5.0/5.1 and mariadb 5.1. - Setup Buildbot upgrade test from MariaDB 5.1.42 ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): Update packaging scripts for MariaDB 5.2 (88)
by worklog-noreply＠askmonty.org 13 Mar '10

13 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Update packaging scripts for MariaDB 5.2 CREATION DATE..: Sat, 27 Feb 2010, 16:39 SUPERVISOR.....: Knielsen IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 88 (http://askmonty.org/worklog/?tid=88) VERSION........: Server-5.2 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 30 (hours remain) ORIG. ESTIMATE.: 30 PROGRESS NOTES: -=-=(Guest - Sat, 13 Mar 2010, 08:12)=-=- Category updated. --- /tmp/wklog.88.old.22167 2010-03-13 08:12:01.000000000 +0000 +++ /tmp/wklog.88.new.22167 2010-03-13 08:12:01.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint DESCRIPTION: The packaging scripts need to be updated to work for MariaDB 5.2 Currently, 5.2 package builds fail in Buildbot. The .debs are missing a debian-5.2 subdirectory. The .rpm also need to be checked. Buildbot needs to be updated to do the new upgrade tests (mariadb-5.1 -> mariadb 5.2) ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 2832: 1. don't crash on failing to load a plugin with newer MYSQL_PLUGIN_INTERFACE_VERSION
by noreply＠launchpad.net 12 Mar '10

12 Mar '10

------------------------------------------------------------ revno: 2832 committer: Sergei Golubchik <sergii(a)pisem.net> branch nick: maria-5.1 timestamp: Fri 2010-03-12 20:05:21 +0100 message: 1. don't crash on failing to load a plugin with newer MYSQL_PLUGIN_INTERFACE_VERSION 2. don't copy st_mysql_plugin structure unnecessary (sizeof hasn't changed) modified: sql/sql_plugin.cc sql/sql_plugin.h -- lp:maria https://code.launchpad.net/~maria-captains/maria/5.1 Your team Maria developers is subscribed to branch lp:maria. To unsubscribe from this branch go to https://code.launchpad.net/~maria-captains/maria/5.1/+edit-subscription.

1 0

[Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 2831: Fix myisam checksum patch to check for HA_OPTION_CHECKSUM after it was set, not before
by noreply＠launchpad.net 12 Mar '10

12 Mar '10

------------------------------------------------------------ revno: 2831 committer: Sergei Golubchik <sergii(a)pisem.net> branch nick: maria-5.1 timestamp: Fri 2010-03-12 20:03:37 +0100 message: Fix myisam checksum patch to check for HA_OPTION_CHECKSUM after it was set, not before modified: storage/myisam/mi_create.c -- lp:maria https://code.launchpad.net/~maria-captains/maria/5.1 Your team Maria developers is subscribed to branch lp:maria. To unsubscribe from this branch go to https://code.launchpad.net/~maria-captains/maria/5.1/+edit-subscription.

1 0

[Maria-developers] Updated (by Timour): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Timour - Fri, 12 Mar 2010, 09:17)=-=- Status updated. --- /tmp/wklog.89.old.13018 2010-03-12 09:17:25.000000000 +0000 +++ /tmp/wklog.89.new.13018 2010-03-12 09:17:25.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Timour): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Timour - Fri, 12 Mar 2010, 09:17)=-=- Status updated. --- /tmp/wklog.89.old.13018 2010-03-12 09:17:25.000000000 +0000 +++ /tmp/wklog.89.new.13018 2010-03-12 09:17:25.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Timour): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Timour - Fri, 12 Mar 2010, 09:17)=-=- Status updated. --- /tmp/wklog.89.old.13018 2010-03-12 09:17:25.000000000 +0000 +++ /tmp/wklog.89.new.13018 2010-03-12 09:17:25.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Timour): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Timour - Fri, 12 Mar 2010, 09:17)=-=- Status updated. --- /tmp/wklog.89.old.13018 2010-03-12 09:17:25.000000000 +0000 +++ /tmp/wklog.89.new.13018 2010-03-12 09:17:25.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Timour): Subquery optimization: Efficient NOT IN execution with NULLs (68)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Efficient NOT IN execution with NULLs CREATION DATE..: Fri, 27 Nov 2009, 13:22 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 68 (http://askmonty.org/worklog/?tid=68) VERSION........: Server-9.x STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 68 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 68 -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 68 -=-=(Guest - Sat, 27 Feb 2010, 10:11)=-=- Status updated. No change. -=-=(Guest - Sat, 27 Feb 2010, 10:11)=-=- Status updated. --- /tmp/wklog.68.old.24229 2010-02-27 10:11:57.000000000 +0000 +++ /tmp/wklog.68.new.24229 2010-02-27 10:11:57.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Timour - Mon, 22 Feb 2010, 17:39)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.17116 2010-02-22 17:39:48.000000000 +0200 +++ /tmp/wklog.68.new.17116 2010-02-22 17:39:48.000000000 +0200 @@ -233,6 +233,7 @@ 1. If columns a_j1,...,a_jm do not contain null values in the temporary table at all and v_j1,...,v_jm cannot be null, create for these columns only one index array (and of course do not create any bitmaps for them). +[done] 2. Consider the ratio d(a_i)=N'(a_i)/V(a_i), where N'(a_i) is the number of rows, where a_i is not null and V(a_i) is the number of distinct @@ -264,6 +265,10 @@ 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. +[This is wrong, because if we don't fill the whole temp table, there may + be some tuple(s) that would match some outer tuple. In such cases, if we + stop filling the temp table, we would miss a TRUE result. Having a partial + match doesn't preclude us from having a complete match]. 8. [timour] Consider that due to materialization, we already have a unique index -=-=(Timour - Tue, 19 Jan 2010, 18:44)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.22569 2010-01-19 18:44:01.000000000 +0200 +++ /tmp/wklog.68.new.22569 2010-01-19 18:44:01.000000000 +0200 @@ -132,11 +132,10 @@ if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE - if (nonull_key) - pq.insert(nonull_key) for (i = 1; i <= n; i++) { + if (vkey[i] != nonull_key) vkey[i].lookup(outer_ref) if (! vkey[i].is_eof()) pq.insert(i) @@ -167,7 +166,7 @@ /* There cannot be a complete match, as we already checked for one. */ assert(matching_keys.elements < n) } - else if (cur_min_key == nonull_key) + else if (vkey[cur_min_key] == nonull_key) { /* The non-NULL key has no corresponding NULL index, so we know for @@ -183,8 +182,10 @@ /* Check if all null_keys contain a NULL at row 'min_row'. The procedure internally checks all keys in a special precomputed order. A prior - procedure determines an optimal order and a mapping - idx_no -> idx_order (encoded as an array). + procedure determines an optimal order and a mapping idx_no -> idx_order + (encoded as an array). + + This procedure makes sure not to match the non-NULL column. */ if (test_null_row(null_keys, min_row)) return TRUE @@ -198,6 +199,14 @@ vkey[cur_min_key].next() if (! vkey[cur_min_key].is_eof()) pq.insert(cur_min_key) + else if (vkey[cur_min_key] == nonull_key) + { + /* + If there can't be more matches for the nonull_key, we know for sure + there is no match, since there is no possible NULL match. + */ + return FALSE + } if (pq.is_empty()) { @@ -216,7 +225,6 @@ } - 3. Directions for improvement ======================================================================== -=-=(Timour - Tue, 19 Jan 2010, 18:29)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.21045 2010-01-19 18:29:12.000000000 +0200 +++ /tmp/wklog.68.new.21045 2010-01-19 18:29:12.000000000 +0200 @@ -132,6 +132,8 @@ if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE + if (nonull_key) + pq.insert(nonull_key) for (i = 1; i <= n; i++) { -=-=(Guest - Tue, 19 Jan 2010, 18:15)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.19825 2010-01-19 18:15:30.000000000 +0200 +++ /tmp/wklog.68.new.19825 2010-01-19 18:15:30.000000000 +0200 @@ -1,8 +1,16 @@ -This a copy of the initial algorithm proposed by Igor: -====================================================== +Contents +======================================================================== -For each left side tuple (v_1,...,v_n) we have to find the following set -of rowids for the temp table containing N rows as the result of +1. Initial idea as proposed by Igor +2. Algorithm for IN execution with partial matching +3. Directions for improvement + + +1. Initial idea as proposed by Igor +======================================================================== + +For each left side tuple (v_1,...,v_n) we have to find the following +set of rowids for the temp table containing N rows as the result of materialization of the subquery: R= INTERSECT (rowid{a_i=v_i} UNION rowid{a_i is null} where i runs @@ -18,38 +26,198 @@ - it requires minimum memory: not more than N*n bits in total - search of an element in a set is extremely cheap -Taken all above into account I could suggest the following algorithm to -build R: +Taken all above into account I could suggest the following algorithm +to build R: - Using indexes (read about them below) for each column participating in the - intersection, - merge ordered sets rowid{a_i=v_i} in the following manner. + Using indexes (read about them below) for each column participating + in the intersection, merge ordered sets rowid{a_i=v_i} in the + following manner. If a rowid r has been encountered maximum in k sets -rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), + rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), then it has to be checked against all rowid{a_i=v_i} such that i is -not in {i1,...,ik}. + not in {i1,...,ik}. As soon as we fail to find r in one of these sets we discard it. If r has been found in all of them then r belongs to the set R. -Here we use the property (1): any r from rowid{a_i=v_i} UNION rowid{a_i -is null} is either +Here we use the property (1): +any r from rowid{a_i=v_i} UNION rowid{a_i is null} is either belongs to rowid{a_i=v_i} or to rowid{a_i is null}. From this we can -infer that for any r from R -indexes a_i can be uniquely divided into two groups: one contains -indexes a_i where r belongs to -the sets rowid{a_i=v_i}, the other contains indexes a_j such that r -belongs to rowid{a_j is null}. - -Now let's talk how to get elements from rowid{a_i=v_i} in a sorted order -needed for the merge procedure. We could use BTREE indexes for temp -table. But they are rather expensive and -take a lot of memory as the are implemented with RB trees. +infer that for any r from R indexes a_i can be uniquely divided into +two groups: +- one contains indexes a_i where r belongs to the sets rowid{a_i=v_i}, +- the other contains indexes a_j such that r belongs to + rowid{a_j is null}. + +Now let's talk how to get elements from rowid{a_i=v_i} in a sorted +order needed for the merge procedure. We could use BTREE indexes for +temp table. But they are rather expensive and take a lot of memory as +the are implemented with RB trees. I would suggest creating for each column from the temporary table just an array of rowids sorted by the value from column a. Index lookup in such an array is cheap. It's also rather cheap to check that the next rowid refers to a row with a different value in column a. The array can be created on demand. +2. Algorithm for IN execution with partial matching +======================================================================== + +2.1 Below is shown the top-level algorithm to execute an IN predicate +with partial matching. This algorithm is essentially the implementation +of Item_subselect:exec(). + +int lookup_with_null_semantics(outer_ref[], mat_subquery) +{ + if (index_lookup(outer_ref, mat_subquery) + return TRUE + else + { + /* + Check if there is a partial match (UNKNOWN) or no match (NULL). + */ + if (this is the first partial match) + { + vkey[] = build array of value keys for each NULL-able column + of mat_subquery. + nkey[] = build a bitmap NULL index for each column of mat_subquery + that contains NULLs + nonull_key = build a key over all non-NULL columns of mat_subquery + } + if (partial_match(outer_ref, vkey[], nkey[], nonull_key) + return UNKNOWN + else + return FALSE + } +} + +2.2 The implementation of partial matching is as follows + +/* + Assumptions: + - It has already been checked if there is a complete match by a + regular index lookup, and the test failed. + - It has already been checked if there is a complete NULL row, + and if there was we wouldn't call this function. Thus we assume + that there is no complete NULL row. + - Not all vidx_i are empty, but some can be empty. If all were empty, + then the only possibility for a match is a complete NULL row, which + we already checked. + + @param outer_ref - the uter (left) IN argument. + @param vidx[] - array of value keys + Ordered sequences of rowids of the corresponding columns a_i, such + that all rowids in idx_i are the ones where column a_i contains some + value or NULL. Each idx_i is derived dynamically, for each different + left argument of an IN predicate. + @param nidx[] - array of NULL keys + Bitmpas, one per each column, where a bit is set if the corresponding + row has a NULL value for the corresponding column. + @nonull_key - the only key over all columns of the materialized subquery + that do not contain NULLs + + @returns + @retval FALSE if there is no match + @retval TRUE if there is a partial match +*/ + +Boolean partial_match(outer_ref, vkey[], nkey[], nonull_key) +{ + /* Set of the keys (columns) that form a partial match. */ + Set matching_keys = {} + /* A subset of all keys that need to be checked for NULL matches. */ + Set null_keys = {} + Int min_key /* Key that contains the current minimum position. */ + Int min_row /* Current row number of min_key. */ + Int cur_min_key, cur_min_row + PriorityQueue pq + + if (nonull_key && ! nonull_key->lookup(outer_ref)) + return FALSE + + for (i = 1; i <= n; i++) + { + vkey[i].lookup(outer_ref) + if (! vkey[i].is_eof()) + pq.insert(i) + } + /* + Not all value keys are empty, thus we don't have only NULL + keys. If we had, the only possible match is a NULL row, and + we cheked there is no such row, therefore the result is known + to be FALSE. + In fact this algorithm makes sense for at least two non-NULL + columns. + */ + assert(pq.elements > 1) + + (min_key, min_row) = pq.pop() + matching_keys.add(min_key) + vkey[min_key].next() + if (! vkey[min_key].is_eof()) + pq.insert(min_key) + + while (TRUE) + { + (cur_min_key, cur_min_row) = pq.pop() + + if (cur_min_row == min_row) + { + matching_keys.add(cur_min_key) + /* There cannot be a complete match, as we already checked for one. */ + assert(matching_keys.elements < n) + } + else if (cur_min_key == nonull_key) + { + /* + The non-NULL key has no corresponding NULL index, so we know for + sure that the row 'min_row' is not a match. + */ + (min_key, min_row) = (cur_min_key, cur_min_row) + matching_keys = {min_key} + } + else + { + assert(cur_min_row > min_row) /* Follows from the use of PQ. */ + null_keys = set_difference(all keys vkey[], matching_keys) + /* + Check if all null_keys contain a NULL at row 'min_row'. The procedure + internally checks all keys in a special precomputed order. A prior + procedure determines an optimal order and a mapping + idx_no -> idx_order (encoded as an array). + */ + if (test_null_row(null_keys, min_row)) + return TRUE + else + { + (min_key, min_row) = (cur_min_key, cur_min_row) + matching_keys = {min_key} + } + } + + vkey[cur_min_key].next() + if (! vkey[cur_min_key].is_eof()) + pq.insert(cur_min_key) + + if (pq.is_empty()) + { + /* Check the last row of the last column in PQ for NULL matches. */ + null_keys = set_difference(all keys vkey[], matching_keys) + if (test_null_row(null_keys, min_row)) + return TRUE + else + return FALSE + } + } + + /* We should never get here. */ + assert(FALSE) + return FALSE +} + + + +3. Directions for improvement +======================================================================== + Other consideration that may be taken into account: 1. If columns a_j1,...,a_jm do not contain null values in the temporary -=-=(Timour - Sun, 06 Dec 2009, 14:36)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.12919 2009-12-06 14:36:18.000000000 +0200 +++ /tmp/wklog.68.new.12919 2009-12-06 14:36:18.000000000 +0200 @@ -87,3 +87,8 @@ 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. +8. [timour] + Consider that due to materialization, we already have a unique index +on all columns <a_1,..., a_n>. We can use the first key part of this index +over column a_1, instead of the index rowid{a_i=v_i}. Thus we can avoid +creating the index rowid{a_i=v_i}. ------------------------------------------------------------ -=-=(View All Progress Notes, 16 total)=-=- http://askmonty.org/worklog/index.pl?tid=68&nolimit=1 DESCRIPTION: The goal of this task is to implement efficient execution of NOT IN subquery predicates of the form: <oe_1,...,oe_n> NOT IN <non_correlated subquery> when either some oe_i, or some subqury result column contains NULLs. The problem with such predicates is that it is possible to use index lookups only when neither argument of the predicate contains NULLs. If some argument contains a NULL, then due to NULL semantics, it plays the role of a wildcard. If we were to use regular index lookups, then we would get 'no match' for some outer tuple (thus the predicate evaluates to FALSE), while the SQL semantics means 'partial match', and the predicate should evaluate to NULL. This task implements an efficient algorithm to compute such 'parial matches', where a NULL matches any value. HIGH-LEVEL SPECIFICATION: Contents ======================================================================== 1. Initial idea as proposed by Igor 2. Algorithm for IN execution with partial matching 3. Directions for improvement 1. Initial idea as proposed by Igor ======================================================================== For each left side tuple (v_1,...,v_n) we have to find the following set of rowids for the temp table containing N rows as the result of materialization of the subquery: R= INTERSECT (rowid{a_i=v_i} UNION rowid{a_i is null} where i runs trough all indexes from [1..n] such that v_i is not null. Bear in mind the following specifics of this intersection: (1) For each i: rowid{a_i=v_i} and rowid{a_i is null} are disjoint (2) For each i: rowid{a_i is null} is the same for each tuple, that is, this set is independent of the left-side tuples. Due to (2) it makes sense to build rowid{a_i is null} only once. A good representation for such sets would be bitmaps: - it requires minimum memory: not more than N*n bits in total - search of an element in a set is extremely cheap Taken all above into account I could suggest the following algorithm to build R: Using indexes (read about them below) for each column participating in the intersection, merge ordered sets rowid{a_i=v_i} in the following manner. If a rowid r has been encountered maximum in k sets rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), then it has to be checked against all rowid{a_i=v_i} such that i is not in {i1,...,ik}. As soon as we fail to find r in one of these sets we discard it. If r has been found in all of them then r belongs to the set R. Here we use the property (1): any r from rowid{a_i=v_i} UNION rowid{a_i is null} is either belongs to rowid{a_i=v_i} or to rowid{a_i is null}. From this we can infer that for any r from R indexes a_i can be uniquely divided into two groups: - one contains indexes a_i where r belongs to the sets rowid{a_i=v_i}, - the other contains indexes a_j such that r belongs to rowid{a_j is null}. Now let's talk how to get elements from rowid{a_i=v_i} in a sorted order needed for the merge procedure. We could use BTREE indexes for temp table. But they are rather expensive and take a lot of memory as the are implemented with RB trees. I would suggest creating for each column from the temporary table just an array of rowids sorted by the value from column a. Index lookup in such an array is cheap. It's also rather cheap to check that the next rowid refers to a row with a different value in column a. The array can be created on demand. 2. Algorithm for IN execution with partial matching ======================================================================== 2.1 Below is shown the top-level algorithm to execute an IN predicate with partial matching. This algorithm is essentially the implementation of Item_subselect:exec(). int lookup_with_null_semantics(outer_ref[], mat_subquery) { if (index_lookup(outer_ref, mat_subquery) return TRUE else { /* Check if there is a partial match (UNKNOWN) or no match (NULL). */ if (this is the first partial match) { vkey[] = build array of value keys for each NULL-able column of mat_subquery. nkey[] = build a bitmap NULL index for each column of mat_subquery that contains NULLs nonull_key = build a key over all non-NULL columns of mat_subquery } if (partial_match(outer_ref, vkey[], nkey[], nonull_key) return UNKNOWN else return FALSE } } 2.2 The implementation of partial matching is as follows /* Assumptions: - It has already been checked if there is a complete match by a regular index lookup, and the test failed. - It has already been checked if there is a complete NULL row, and if there was we wouldn't call this function. Thus we assume that there is no complete NULL row. - Not all vidx_i are empty, but some can be empty. If all were empty, then the only possibility for a match is a complete NULL row, which we already checked. @param outer_ref - the uter (left) IN argument. @param vidx[] - array of value keys Ordered sequences of rowids of the corresponding columns a_i, such that all rowids in idx_i are the ones where column a_i contains some value or NULL. Each idx_i is derived dynamically, for each different left argument of an IN predicate. @param nidx[] - array of NULL keys Bitmpas, one per each column, where a bit is set if the corresponding row has a NULL value for the corresponding column. @nonull_key - the only key over all columns of the materialized subquery that do not contain NULLs @returns @retval FALSE if there is no match @retval TRUE if there is a partial match */ Boolean partial_match(outer_ref, vkey[], nkey[], nonull_key) { /* Set of the keys (columns) that form a partial match. */ Set matching_keys = {} /* A subset of all keys that need to be checked for NULL matches. */ Set null_keys = {} Int min_key /* Key that contains the current minimum position. */ Int min_row /* Current row number of min_key. */ Int cur_min_key, cur_min_row PriorityQueue pq if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE for (i = 1; i <= n; i++) { if (vkey[i] != nonull_key) vkey[i].lookup(outer_ref) if (! vkey[i].is_eof()) pq.insert(i) } /* Not all value keys are empty, thus we don't have only NULL keys. If we had, the only possible match is a NULL row, and we cheked there is no such row, therefore the result is known to be FALSE. In fact this algorithm makes sense for at least two non-NULL columns. */ assert(pq.elements > 1) (min_key, min_row) = pq.pop() matching_keys.add(min_key) vkey[min_key].next() if (! vkey[min_key].is_eof()) pq.insert(min_key) while (TRUE) { (cur_min_key, cur_min_row) = pq.pop() if (cur_min_row == min_row) { matching_keys.add(cur_min_key) /* There cannot be a complete match, as we already checked for one. */ assert(matching_keys.elements < n) } else if (vkey[cur_min_key] == nonull_key) { /* The non-NULL key has no corresponding NULL index, so we know for sure that the row 'min_row' is not a match. */ (min_key, min_row) = (cur_min_key, cur_min_row) matching_keys = {min_key} } else { assert(cur_min_row > min_row) /* Follows from the use of PQ. */ null_keys = set_difference(all keys vkey[], matching_keys) /* Check if all null_keys contain a NULL at row 'min_row'. The procedure internally checks all keys in a special precomputed order. A prior procedure determines an optimal order and a mapping idx_no -> idx_order (encoded as an array). This procedure makes sure not to match the non-NULL column. */ if (test_null_row(null_keys, min_row)) return TRUE else { (min_key, min_row) = (cur_min_key, cur_min_row) matching_keys = {min_key} } } vkey[cur_min_key].next() if (! vkey[cur_min_key].is_eof()) pq.insert(cur_min_key) else if (vkey[cur_min_key] == nonull_key) { /* If there can't be more matches for the nonull_key, we know for sure there is no match, since there is no possible NULL match. */ return FALSE } if (pq.is_empty()) { /* Check the last row of the last column in PQ for NULL matches. */ null_keys = set_difference(all keys vkey[], matching_keys) if (test_null_row(null_keys, min_row)) return TRUE else return FALSE } } /* We should never get here. */ assert(FALSE) return FALSE } 3. Directions for improvement ======================================================================== Other consideration that may be taken into account: 1. If columns a_j1,...,a_jm do not contain null values in the temporary table at all and v_j1,...,v_jm cannot be null, create for these columns only one index array (and of course do not create any bitmaps for them). [done] 2. Consider the ratio d(a_i)=N'(a_i)/V(a_i), where N'(a_i) is the number of rows, where a_i is not null and V(a_i) is the number of distinct values for a_i excluding nulls. If d(a_i) is close to N'(a_i) then do not create any index array: check whether there is a match running through the records that have been filtered in. Anyway if d(a_i) is close to N'(a_i) then the intersection with rowid{a_i=v_i} will not reduce the number of remaining rowids significantly. In other words is V(a_i) exceeds some threshold there is no sense to create an index for a_i. If additionally N-N'(a_i) is small do not create a bitmap for this column either. 3. If for a column a_i d(a_i) is not close to N'(a_i), but N-N'(a_i) is small a sorted array of rowids from the set rowid{a_i is null} can be used instead of a bitmap. 4. We always have a match if R0= INTERSECT rowid{a_i is null} is not empty. Here i runs through all indexes from [1..n] such that v_i is not null. For a given subset of columns this fact has to be checked only once. It can be easily done with bitmap intersection. 5. If v1,...,vn never can be a null, then indexes (sorted arrays) can be created only for rows with nulls. 6. If v1,...,vn never can be a null and number of rows with nulls is small do not create indexes and do not create bitmaps. 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. [This is wrong, because if we don't fill the whole temp table, there may be some tuple(s) that would match some outer tuple. In such cases, if we stop filling the temp table, we would miss a TRUE result. Having a partial match doesn't preclude us from having a complete match]. 8. [timour] Consider that due to materialization, we already have a unique index on all columns <a_1,..., a_n>. We can use the first key part of this index over column a_1, instead of the index rowid{a_i=v_i}. Thus we can avoid creating the index rowid{a_i=v_i}. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Timour): Subquery optimization: Efficient NOT IN execution with NULLs (68)
by worklog-noreply＠askmonty.org 12 Mar '10

12 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Efficient NOT IN execution with NULLs CREATION DATE..: Fri, 27 Nov 2009, 13:22 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 68 (http://askmonty.org/worklog/?tid=68) VERSION........: Server-9.x STATUS.........: In-Progress PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 68 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 68 -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 68 -=-=(Guest - Sat, 27 Feb 2010, 10:11)=-=- Status updated. No change. -=-=(Guest - Sat, 27 Feb 2010, 10:11)=-=- Status updated. --- /tmp/wklog.68.old.24229 2010-02-27 10:11:57.000000000 +0000 +++ /tmp/wklog.68.new.24229 2010-02-27 10:11:57.000000000 +0000 @@ -1 +1 @@ -Assigned +In-Progress -=-=(Timour - Mon, 22 Feb 2010, 17:39)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.17116 2010-02-22 17:39:48.000000000 +0200 +++ /tmp/wklog.68.new.17116 2010-02-22 17:39:48.000000000 +0200 @@ -233,6 +233,7 @@ 1. If columns a_j1,...,a_jm do not contain null values in the temporary table at all and v_j1,...,v_jm cannot be null, create for these columns only one index array (and of course do not create any bitmaps for them). +[done] 2. Consider the ratio d(a_i)=N'(a_i)/V(a_i), where N'(a_i) is the number of rows, where a_i is not null and V(a_i) is the number of distinct @@ -264,6 +265,10 @@ 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. +[This is wrong, because if we don't fill the whole temp table, there may + be some tuple(s) that would match some outer tuple. In such cases, if we + stop filling the temp table, we would miss a TRUE result. Having a partial + match doesn't preclude us from having a complete match]. 8. [timour] Consider that due to materialization, we already have a unique index -=-=(Timour - Tue, 19 Jan 2010, 18:44)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.22569 2010-01-19 18:44:01.000000000 +0200 +++ /tmp/wklog.68.new.22569 2010-01-19 18:44:01.000000000 +0200 @@ -132,11 +132,10 @@ if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE - if (nonull_key) - pq.insert(nonull_key) for (i = 1; i <= n; i++) { + if (vkey[i] != nonull_key) vkey[i].lookup(outer_ref) if (! vkey[i].is_eof()) pq.insert(i) @@ -167,7 +166,7 @@ /* There cannot be a complete match, as we already checked for one. */ assert(matching_keys.elements < n) } - else if (cur_min_key == nonull_key) + else if (vkey[cur_min_key] == nonull_key) { /* The non-NULL key has no corresponding NULL index, so we know for @@ -183,8 +182,10 @@ /* Check if all null_keys contain a NULL at row 'min_row'. The procedure internally checks all keys in a special precomputed order. A prior - procedure determines an optimal order and a mapping - idx_no -> idx_order (encoded as an array). + procedure determines an optimal order and a mapping idx_no -> idx_order + (encoded as an array). + + This procedure makes sure not to match the non-NULL column. */ if (test_null_row(null_keys, min_row)) return TRUE @@ -198,6 +199,14 @@ vkey[cur_min_key].next() if (! vkey[cur_min_key].is_eof()) pq.insert(cur_min_key) + else if (vkey[cur_min_key] == nonull_key) + { + /* + If there can't be more matches for the nonull_key, we know for sure + there is no match, since there is no possible NULL match. + */ + return FALSE + } if (pq.is_empty()) { @@ -216,7 +225,6 @@ } - 3. Directions for improvement ======================================================================== -=-=(Timour - Tue, 19 Jan 2010, 18:29)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.21045 2010-01-19 18:29:12.000000000 +0200 +++ /tmp/wklog.68.new.21045 2010-01-19 18:29:12.000000000 +0200 @@ -132,6 +132,8 @@ if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE + if (nonull_key) + pq.insert(nonull_key) for (i = 1; i <= n; i++) { -=-=(Guest - Tue, 19 Jan 2010, 18:15)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.19825 2010-01-19 18:15:30.000000000 +0200 +++ /tmp/wklog.68.new.19825 2010-01-19 18:15:30.000000000 +0200 @@ -1,8 +1,16 @@ -This a copy of the initial algorithm proposed by Igor: -====================================================== +Contents +======================================================================== -For each left side tuple (v_1,...,v_n) we have to find the following set -of rowids for the temp table containing N rows as the result of +1. Initial idea as proposed by Igor +2. Algorithm for IN execution with partial matching +3. Directions for improvement + + +1. Initial idea as proposed by Igor +======================================================================== + +For each left side tuple (v_1,...,v_n) we have to find the following +set of rowids for the temp table containing N rows as the result of materialization of the subquery: R= INTERSECT (rowid{a_i=v_i} UNION rowid{a_i is null} where i runs @@ -18,38 +26,198 @@ - it requires minimum memory: not more than N*n bits in total - search of an element in a set is extremely cheap -Taken all above into account I could suggest the following algorithm to -build R: +Taken all above into account I could suggest the following algorithm +to build R: - Using indexes (read about them below) for each column participating in the - intersection, - merge ordered sets rowid{a_i=v_i} in the following manner. + Using indexes (read about them below) for each column participating + in the intersection, merge ordered sets rowid{a_i=v_i} in the + following manner. If a rowid r has been encountered maximum in k sets -rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), + rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), then it has to be checked against all rowid{a_i=v_i} such that i is -not in {i1,...,ik}. + not in {i1,...,ik}. As soon as we fail to find r in one of these sets we discard it. If r has been found in all of them then r belongs to the set R. -Here we use the property (1): any r from rowid{a_i=v_i} UNION rowid{a_i -is null} is either +Here we use the property (1): +any r from rowid{a_i=v_i} UNION rowid{a_i is null} is either belongs to rowid{a_i=v_i} or to rowid{a_i is null}. From this we can -infer that for any r from R -indexes a_i can be uniquely divided into two groups: one contains -indexes a_i where r belongs to -the sets rowid{a_i=v_i}, the other contains indexes a_j such that r -belongs to rowid{a_j is null}. - -Now let's talk how to get elements from rowid{a_i=v_i} in a sorted order -needed for the merge procedure. We could use BTREE indexes for temp -table. But they are rather expensive and -take a lot of memory as the are implemented with RB trees. +infer that for any r from R indexes a_i can be uniquely divided into +two groups: +- one contains indexes a_i where r belongs to the sets rowid{a_i=v_i}, +- the other contains indexes a_j such that r belongs to + rowid{a_j is null}. + +Now let's talk how to get elements from rowid{a_i=v_i} in a sorted +order needed for the merge procedure. We could use BTREE indexes for +temp table. But they are rather expensive and take a lot of memory as +the are implemented with RB trees. I would suggest creating for each column from the temporary table just an array of rowids sorted by the value from column a. Index lookup in such an array is cheap. It's also rather cheap to check that the next rowid refers to a row with a different value in column a. The array can be created on demand. +2. Algorithm for IN execution with partial matching +======================================================================== + +2.1 Below is shown the top-level algorithm to execute an IN predicate +with partial matching. This algorithm is essentially the implementation +of Item_subselect:exec(). + +int lookup_with_null_semantics(outer_ref[], mat_subquery) +{ + if (index_lookup(outer_ref, mat_subquery) + return TRUE + else + { + /* + Check if there is a partial match (UNKNOWN) or no match (NULL). + */ + if (this is the first partial match) + { + vkey[] = build array of value keys for each NULL-able column + of mat_subquery. + nkey[] = build a bitmap NULL index for each column of mat_subquery + that contains NULLs + nonull_key = build a key over all non-NULL columns of mat_subquery + } + if (partial_match(outer_ref, vkey[], nkey[], nonull_key) + return UNKNOWN + else + return FALSE + } +} + +2.2 The implementation of partial matching is as follows + +/* + Assumptions: + - It has already been checked if there is a complete match by a + regular index lookup, and the test failed. + - It has already been checked if there is a complete NULL row, + and if there was we wouldn't call this function. Thus we assume + that there is no complete NULL row. + - Not all vidx_i are empty, but some can be empty. If all were empty, + then the only possibility for a match is a complete NULL row, which + we already checked. + + @param outer_ref - the uter (left) IN argument. + @param vidx[] - array of value keys + Ordered sequences of rowids of the corresponding columns a_i, such + that all rowids in idx_i are the ones where column a_i contains some + value or NULL. Each idx_i is derived dynamically, for each different + left argument of an IN predicate. + @param nidx[] - array of NULL keys + Bitmpas, one per each column, where a bit is set if the corresponding + row has a NULL value for the corresponding column. + @nonull_key - the only key over all columns of the materialized subquery + that do not contain NULLs + + @returns + @retval FALSE if there is no match + @retval TRUE if there is a partial match +*/ + +Boolean partial_match(outer_ref, vkey[], nkey[], nonull_key) +{ + /* Set of the keys (columns) that form a partial match. */ + Set matching_keys = {} + /* A subset of all keys that need to be checked for NULL matches. */ + Set null_keys = {} + Int min_key /* Key that contains the current minimum position. */ + Int min_row /* Current row number of min_key. */ + Int cur_min_key, cur_min_row + PriorityQueue pq + + if (nonull_key && ! nonull_key->lookup(outer_ref)) + return FALSE + + for (i = 1; i <= n; i++) + { + vkey[i].lookup(outer_ref) + if (! vkey[i].is_eof()) + pq.insert(i) + } + /* + Not all value keys are empty, thus we don't have only NULL + keys. If we had, the only possible match is a NULL row, and + we cheked there is no such row, therefore the result is known + to be FALSE. + In fact this algorithm makes sense for at least two non-NULL + columns. + */ + assert(pq.elements > 1) + + (min_key, min_row) = pq.pop() + matching_keys.add(min_key) + vkey[min_key].next() + if (! vkey[min_key].is_eof()) + pq.insert(min_key) + + while (TRUE) + { + (cur_min_key, cur_min_row) = pq.pop() + + if (cur_min_row == min_row) + { + matching_keys.add(cur_min_key) + /* There cannot be a complete match, as we already checked for one. */ + assert(matching_keys.elements < n) + } + else if (cur_min_key == nonull_key) + { + /* + The non-NULL key has no corresponding NULL index, so we know for + sure that the row 'min_row' is not a match. + */ + (min_key, min_row) = (cur_min_key, cur_min_row) + matching_keys = {min_key} + } + else + { + assert(cur_min_row > min_row) /* Follows from the use of PQ. */ + null_keys = set_difference(all keys vkey[], matching_keys) + /* + Check if all null_keys contain a NULL at row 'min_row'. The procedure + internally checks all keys in a special precomputed order. A prior + procedure determines an optimal order and a mapping + idx_no -> idx_order (encoded as an array). + */ + if (test_null_row(null_keys, min_row)) + return TRUE + else + { + (min_key, min_row) = (cur_min_key, cur_min_row) + matching_keys = {min_key} + } + } + + vkey[cur_min_key].next() + if (! vkey[cur_min_key].is_eof()) + pq.insert(cur_min_key) + + if (pq.is_empty()) + { + /* Check the last row of the last column in PQ for NULL matches. */ + null_keys = set_difference(all keys vkey[], matching_keys) + if (test_null_row(null_keys, min_row)) + return TRUE + else + return FALSE + } + } + + /* We should never get here. */ + assert(FALSE) + return FALSE +} + + + +3. Directions for improvement +======================================================================== + Other consideration that may be taken into account: 1. If columns a_j1,...,a_jm do not contain null values in the temporary -=-=(Timour - Sun, 06 Dec 2009, 14:36)=-=- High-Level Specification modified. --- /tmp/wklog.68.old.12919 2009-12-06 14:36:18.000000000 +0200 +++ /tmp/wklog.68.new.12919 2009-12-06 14:36:18.000000000 +0200 @@ -87,3 +87,8 @@ 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. +8. [timour] + Consider that due to materialization, we already have a unique index +on all columns <a_1,..., a_n>. We can use the first key part of this index +over column a_1, instead of the index rowid{a_i=v_i}. Thus we can avoid +creating the index rowid{a_i=v_i}. ------------------------------------------------------------ -=-=(View All Progress Notes, 16 total)=-=- http://askmonty.org/worklog/index.pl?tid=68&nolimit=1 DESCRIPTION: The goal of this task is to implement efficient execution of NOT IN subquery predicates of the form: <oe_1,...,oe_n> NOT IN <non_correlated subquery> when either some oe_i, or some subqury result column contains NULLs. The problem with such predicates is that it is possible to use index lookups only when neither argument of the predicate contains NULLs. If some argument contains a NULL, then due to NULL semantics, it plays the role of a wildcard. If we were to use regular index lookups, then we would get 'no match' for some outer tuple (thus the predicate evaluates to FALSE), while the SQL semantics means 'partial match', and the predicate should evaluate to NULL. This task implements an efficient algorithm to compute such 'parial matches', where a NULL matches any value. HIGH-LEVEL SPECIFICATION: Contents ======================================================================== 1. Initial idea as proposed by Igor 2. Algorithm for IN execution with partial matching 3. Directions for improvement 1. Initial idea as proposed by Igor ======================================================================== For each left side tuple (v_1,...,v_n) we have to find the following set of rowids for the temp table containing N rows as the result of materialization of the subquery: R= INTERSECT (rowid{a_i=v_i} UNION rowid{a_i is null} where i runs trough all indexes from [1..n] such that v_i is not null. Bear in mind the following specifics of this intersection: (1) For each i: rowid{a_i=v_i} and rowid{a_i is null} are disjoint (2) For each i: rowid{a_i is null} is the same for each tuple, that is, this set is independent of the left-side tuples. Due to (2) it makes sense to build rowid{a_i is null} only once. A good representation for such sets would be bitmaps: - it requires minimum memory: not more than N*n bits in total - search of an element in a set is extremely cheap Taken all above into account I could suggest the following algorithm to build R: Using indexes (read about them below) for each column participating in the intersection, merge ordered sets rowid{a_i=v_i} in the following manner. If a rowid r has been encountered maximum in k sets rowid{a_i1=v_i1},...,rowid(a_ik=v_ik), then it has to be checked against all rowid{a_i=v_i} such that i is not in {i1,...,ik}. As soon as we fail to find r in one of these sets we discard it. If r has been found in all of them then r belongs to the set R. Here we use the property (1): any r from rowid{a_i=v_i} UNION rowid{a_i is null} is either belongs to rowid{a_i=v_i} or to rowid{a_i is null}. From this we can infer that for any r from R indexes a_i can be uniquely divided into two groups: - one contains indexes a_i where r belongs to the sets rowid{a_i=v_i}, - the other contains indexes a_j such that r belongs to rowid{a_j is null}. Now let's talk how to get elements from rowid{a_i=v_i} in a sorted order needed for the merge procedure. We could use BTREE indexes for temp table. But they are rather expensive and take a lot of memory as the are implemented with RB trees. I would suggest creating for each column from the temporary table just an array of rowids sorted by the value from column a. Index lookup in such an array is cheap. It's also rather cheap to check that the next rowid refers to a row with a different value in column a. The array can be created on demand. 2. Algorithm for IN execution with partial matching ======================================================================== 2.1 Below is shown the top-level algorithm to execute an IN predicate with partial matching. This algorithm is essentially the implementation of Item_subselect:exec(). int lookup_with_null_semantics(outer_ref[], mat_subquery) { if (index_lookup(outer_ref, mat_subquery) return TRUE else { /* Check if there is a partial match (UNKNOWN) or no match (NULL). */ if (this is the first partial match) { vkey[] = build array of value keys for each NULL-able column of mat_subquery. nkey[] = build a bitmap NULL index for each column of mat_subquery that contains NULLs nonull_key = build a key over all non-NULL columns of mat_subquery } if (partial_match(outer_ref, vkey[], nkey[], nonull_key) return UNKNOWN else return FALSE } } 2.2 The implementation of partial matching is as follows /* Assumptions: - It has already been checked if there is a complete match by a regular index lookup, and the test failed. - It has already been checked if there is a complete NULL row, and if there was we wouldn't call this function. Thus we assume that there is no complete NULL row. - Not all vidx_i are empty, but some can be empty. If all were empty, then the only possibility for a match is a complete NULL row, which we already checked. @param outer_ref - the uter (left) IN argument. @param vidx[] - array of value keys Ordered sequences of rowids of the corresponding columns a_i, such that all rowids in idx_i are the ones where column a_i contains some value or NULL. Each idx_i is derived dynamically, for each different left argument of an IN predicate. @param nidx[] - array of NULL keys Bitmpas, one per each column, where a bit is set if the corresponding row has a NULL value for the corresponding column. @nonull_key - the only key over all columns of the materialized subquery that do not contain NULLs @returns @retval FALSE if there is no match @retval TRUE if there is a partial match */ Boolean partial_match(outer_ref, vkey[], nkey[], nonull_key) { /* Set of the keys (columns) that form a partial match. */ Set matching_keys = {} /* A subset of all keys that need to be checked for NULL matches. */ Set null_keys = {} Int min_key /* Key that contains the current minimum position. */ Int min_row /* Current row number of min_key. */ Int cur_min_key, cur_min_row PriorityQueue pq if (nonull_key && ! nonull_key->lookup(outer_ref)) return FALSE for (i = 1; i <= n; i++) { if (vkey[i] != nonull_key) vkey[i].lookup(outer_ref) if (! vkey[i].is_eof()) pq.insert(i) } /* Not all value keys are empty, thus we don't have only NULL keys. If we had, the only possible match is a NULL row, and we cheked there is no such row, therefore the result is known to be FALSE. In fact this algorithm makes sense for at least two non-NULL columns. */ assert(pq.elements > 1) (min_key, min_row) = pq.pop() matching_keys.add(min_key) vkey[min_key].next() if (! vkey[min_key].is_eof()) pq.insert(min_key) while (TRUE) { (cur_min_key, cur_min_row) = pq.pop() if (cur_min_row == min_row) { matching_keys.add(cur_min_key) /* There cannot be a complete match, as we already checked for one. */ assert(matching_keys.elements < n) } else if (vkey[cur_min_key] == nonull_key) { /* The non-NULL key has no corresponding NULL index, so we know for sure that the row 'min_row' is not a match. */ (min_key, min_row) = (cur_min_key, cur_min_row) matching_keys = {min_key} } else { assert(cur_min_row > min_row) /* Follows from the use of PQ. */ null_keys = set_difference(all keys vkey[], matching_keys) /* Check if all null_keys contain a NULL at row 'min_row'. The procedure internally checks all keys in a special precomputed order. A prior procedure determines an optimal order and a mapping idx_no -> idx_order (encoded as an array). This procedure makes sure not to match the non-NULL column. */ if (test_null_row(null_keys, min_row)) return TRUE else { (min_key, min_row) = (cur_min_key, cur_min_row) matching_keys = {min_key} } } vkey[cur_min_key].next() if (! vkey[cur_min_key].is_eof()) pq.insert(cur_min_key) else if (vkey[cur_min_key] == nonull_key) { /* If there can't be more matches for the nonull_key, we know for sure there is no match, since there is no possible NULL match. */ return FALSE } if (pq.is_empty()) { /* Check the last row of the last column in PQ for NULL matches. */ null_keys = set_difference(all keys vkey[], matching_keys) if (test_null_row(null_keys, min_row)) return TRUE else return FALSE } } /* We should never get here. */ assert(FALSE) return FALSE } 3. Directions for improvement ======================================================================== Other consideration that may be taken into account: 1. If columns a_j1,...,a_jm do not contain null values in the temporary table at all and v_j1,...,v_jm cannot be null, create for these columns only one index array (and of course do not create any bitmaps for them). [done] 2. Consider the ratio d(a_i)=N'(a_i)/V(a_i), where N'(a_i) is the number of rows, where a_i is not null and V(a_i) is the number of distinct values for a_i excluding nulls. If d(a_i) is close to N'(a_i) then do not create any index array: check whether there is a match running through the records that have been filtered in. Anyway if d(a_i) is close to N'(a_i) then the intersection with rowid{a_i=v_i} will not reduce the number of remaining rowids significantly. In other words is V(a_i) exceeds some threshold there is no sense to create an index for a_i. If additionally N-N'(a_i) is small do not create a bitmap for this column either. 3. If for a column a_i d(a_i) is not close to N'(a_i), but N-N'(a_i) is small a sorted array of rowids from the set rowid{a_i is null} can be used instead of a bitmap. 4. We always have a match if R0= INTERSECT rowid{a_i is null} is not empty. Here i runs through all indexes from [1..n] such that v_i is not null. For a given subset of columns this fact has to be checked only once. It can be easily done with bitmap intersection. 5. If v1,...,vn never can be a null, then indexes (sorted arrays) can be created only for rows with nulls. 6. If v1,...,vn never can be a null and number of rows with nulls is small do not create indexes and do not create bitmaps. 7. If you get a row with nulls in all columns stop filling the temporary table and return UNKNOWN for any tuple <v1,...,vn>. [This is wrong, because if we don't fill the whole temp table, there may be some tuple(s) that would match some outer tuple. In such cases, if we stop filling the temp table, we would miss a TRUE result. Having a partial match doesn't preclude us from having a complete match]. 8. [timour] Consider that due to materialization, we already have a unique index on all columns <a_1,..., a_n>. We can use the first key part of this index over column a_1, instead of the index rowid{a_i=v_i}. Thus we can avoid creating the index rowid{a_i=v_i}. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Rev 2767: MWL#68 Subquery optimization: Efficient NOT IN execution with NULLs in file:///home/tsk/mprog/src/5.3-mwl68/
by timour＠askmonty.org 11 Mar '10

11 Mar '10

At file:///home/tsk/mprog/src/5.3-mwl68/ ------------------------------------------------------------ revno: 2767 revision-id: timour(a)askmonty.org-20100311214331-kw8ng8aiy6h60vai parent: timour(a)askmonty.org-20100309103615-dzmm6xt7ye5xfs25 committer: timour(a)askmonty.org branch nick: 5.3-mwl68 timestamp: Thu 2010-03-11 23:43:31 +0200 message: MWL#68 Subquery optimization: Efficient NOT IN execution with NULLs This patch does three things: - It adds the possibility to force the execution of top-level [NOT] IN subquery predicates via the IN=>EXISTS transformation. This is done by setting both optimizer switches partial_match_rowid_merge and partial_match_table_scan to "off". - It adjusts all test cases where the complete optimizer_switch is selected because now we have two more switches. - For those test cases where the plan changes because of the new available strategies, we switch off both partial match strategies in order to force the "old" IN=>EXISTS strategy. This is done because most of these test cases specifically test bugs in this strategy. === modified file 'mysql-test/include/mix1.inc' --- a/mysql-test/include/mix1.inc 2009-09-15 06:08:54 +0000 +++ b/mysql-test/include/mix1.inc 2010-03-11 21:43:31 +0000 @@ -1177,8 +1177,11 @@ create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; --echo End of 5.0 tests === modified file 'mysql-test/r/index_merge_myisam.result' --- a/mysql-test/r/index_merge_myisam.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/index_merge_myisam.result 2010-03-11 21:43:31 +0000 @@ -1419,19 +1419,19 @@ # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge=off,index_merge_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge_union=on'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,index_merge_sort_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=4; ERROR 42000: Variable 'optimizer_switch' can't be set to the value of '4' set optimizer_switch=NULL; @@ -1458,21 +1458,21 @@ set optimizer_switch='index_merge=off,index_merge_union=off,default'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set @@global.optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on # # Check index_merge's @@optimizer_switch flags # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, c int, filler char(100), @@ -1582,5 +1582,5 @@ set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on drop table t0, t1; === modified file 'mysql-test/r/innodb_mysql.result' --- a/mysql-test/r/innodb_mysql.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mysql.result 2010-03-11 21:43:31 +0000 @@ -1425,12 +1425,15 @@ # create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables 2 DEPENDENT SUBQUERY t1 system NULL NULL NULL NULL 0 const row not found 2 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 1 +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; End of 5.0 tests CREATE TABLE `t2` ( === modified file 'mysql-test/r/myisam_mrr.result' --- a/mysql-test/r/myisam_mrr.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/myisam_mrr.result 2010-03-11 21:43:31 +0000 @@ -394,7 +394,7 @@ # - engine_condition_pushdown does not affect ICP select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, key(a)); === modified file 'mysql-test/r/ps.result' --- a/mysql-test/r/ps.result 2009-05-27 15:19:44 +0000 +++ b/mysql-test/r/ps.result 2010-03-11 21:43:31 +0000 @@ -149,6 +149,8 @@ c32 set('monday', 'tuesday', 'wednesday') ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -177,6 +179,7 @@ 2 DEPENDENT SUBQUERY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; set @arg00=1; prepare stmt1 from ' create table t1 (m int) as select 1 as m ' ; execute stmt1 ; === modified file 'mysql-test/r/subselect.result' --- a/mysql-test/r/subselect.result 2010-02-17 21:59:41 +0000 +++ b/mysql-test/r/subselect.result 2010-03-11 21:43:31 +0000 @@ -1,4 +1,6 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4803,4 +4805,5 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. === modified file 'mysql-test/r/subselect3.result' --- a/mysql-test/r/subselect3.result 2010-02-17 10:05:27 +0000 +++ b/mysql-test/r/subselect3.result 2010-03-11 21:43:31 +0000 @@ -63,12 +63,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -692,6 +695,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -759,6 +764,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -960,7 +966,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -977,7 +983,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect3_jcl6.result' --- a/mysql-test/r/subselect3_jcl6.result 2010-02-17 10:47:55 +0000 +++ b/mysql-test/r/subselect3_jcl6.result 2010-03-11 21:43:31 +0000 @@ -67,12 +67,15 @@ select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -696,6 +699,8 @@ 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -763,6 +768,7 @@ 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -964,7 +970,7 @@ # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -981,7 +987,7 @@ # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect_no_mat.result' --- a/mysql-test/r/subselect_no_mat.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_mat.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_opts.result' --- a/mysql-test/r/subselect_no_opts.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_opts.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off,semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_semijoin.result' --- a/mysql-test/r/subselect_no_semijoin.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_semijoin.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-02-24 11:33:42 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-11 21:43:31 +0000 @@ -202,39 +202,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-07 15:41:45 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-11 21:43:31 +0000 @@ -206,39 +206,39 @@ select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/t/ps.test' --- a/mysql-test/t/ps.test 2009-05-27 15:19:44 +0000 +++ b/mysql-test/t/ps.test 2010-03-11 21:43:31 +0000 @@ -163,6 +163,9 @@ ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -171,6 +174,8 @@ deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # parameters from variables (for field creation) # === modified file 'mysql-test/t/subselect.test' --- a/mysql-test/t/subselect.test 2010-01-17 20:52:20 +0000 +++ b/mysql-test/t/subselect.test 2010-03-11 21:43:31 +0000 @@ -11,6 +11,9 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; --enable_warnings +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + select (select 2); explain extended select (select 2); SELECT (SELECT 1) UNION SELECT (SELECT 2); @@ -4061,4 +4064,6 @@ (SELECT LAST_INSERT_ID() FROM t1 ORDER BY MIN(a) ASC LIMIT 1); DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; + --echo End of 5.1 tests. === modified file 'mysql-test/t/subselect3.test' --- a/mysql-test/t/subselect3.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect3.test 2010-03-11 21:43:31 +0000 @@ -59,9 +59,13 @@ show status like 'Handler_read_rnd_next'; select ' ^ This must show 11' Z; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + # This must show trigcond: explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; # @@ -529,6 +533,9 @@ DROP TABLE t1, t2; +# The next three test cases must be executed with the IN=>EXISTS strategy +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; # # Bug #27870: crash of an equijoin query with WHERE condition containing @@ -588,6 +595,8 @@ DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #34763: item_subselect.cc:1235:Item_in_subselect::row_value_transformer: # Assertion failed, unexpected error message: === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-09 10:36:15 +0000 +++ b/sql/opt_subselect.cc 2010-03-11 21:43:31 +0000 @@ -187,7 +187,11 @@ does not call setup_subquery_materialization(). We could make SELECT ... FROM DUAL call that function but that doesn't seem to be the case that is worth handling. - 4. Subquery is non-correlated + 4. Either the subquery predicate is a top-level predicate, or at + least one partial match strategy is enabled. If no partial match + strategy is enabled, then materialization cannot be used for + non-top-level queries because it cannot handle NULLs correctly. + 5. Subquery is non-correlated TODO: This is an overly restrictive condition. It can be extended to: (Subquery is non-correlated || @@ -195,13 +199,13 @@ (Subquery is correlated to the immediate outer query && Subquery !contains {GROUP BY, ORDER BY [LIMIT], aggregate functions}) && subquery predicate is not under "NOT IN")) - 5. No execution method was already chosen (by a prepared statement). + 6. No execution method was already chosen (by a prepared statement). (*) The subquery must be part of a SELECT statement. The current condition also excludes multi-table update statements. - We have to determine whether we will perform subquery materialization - before calling the IN=>EXISTS transformation, so that we know whether to + Determine whether we will perform subquery materialization before + calling the IN=>EXISTS transformation, so that we know whether to perform the whole transformation or only that part of it which wraps Item_in_subselect in an Item_in_optimizer. */ @@ -211,11 +215,14 @@ select_lex->master_unit()->first_select()->leaf_tables && // 3 thd->lex->sql_command == SQLCOM_SELECT && // * select_lex->outer_select()->leaf_tables && // 3A - subquery_types_allow_materialization(in_subs)) + subquery_types_allow_materialization(in_subs) && + // psergey-todo: duplicated_subselect_card_check: where it's done? + (in_subs->is_top_level_item() || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) &&//4 + !in_subs->is_correlated && // 5 + in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 { - // psergey-todo: duplicated_subselect_card_check: where it's done? - if (!in_subs->is_correlated && // 4 - in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 5 in_subs->exec_method= Item_in_subselect::MATERIALIZATION; }

1 0

[Maria-developers] bzr commit into file:///home/tsk/mprog/src/5.3-mwl68/ branch (timour:2767)
by timour＠askmonty.org 11 Mar '10

11 Mar '10

#At file:///home/tsk/mprog/src/5.3-mwl68/ based on revid:timour@askmonty.org-20100309103615-dzmm6xt7ye5xfs25 2767 timour(a)askmonty.org 2010-03-11 MWL#68 Subquery optimization: Efficient NOT IN execution with NULLs This patch does three things: - It adds the possibility to force the execution of top-level [NOT] IN subquery predicates via the IN=>EXISTS transformation. This is done by setting both optimizer switches partial_match_rowid_merge and partial_match_table_scan to "off". - It adjusts all test cases where the complete optimizer_switch is selected because now we have two more switches. - For those test cases where the plan changes because of the new available strategies, we switch off both partial match strategies in order to force the "old" IN=>EXISTS strategy. This is done because most of these test cases specifically test bugs in this strategy. @ sql/opt_subselect.cc Adds the possibility to force the execution of top-level [NOT] IN subquery predicates via the IN=>EXISTS transformation. This is done by setting both optimizer switches partial_match_rowid_merge and partial_match_table_scan to "off". modified: mysql-test/include/mix1.inc mysql-test/r/index_merge_myisam.result mysql-test/r/innodb_mysql.result mysql-test/r/myisam_mrr.result mysql-test/r/ps.result mysql-test/r/subselect.result mysql-test/r/subselect3.result mysql-test/r/subselect3_jcl6.result mysql-test/r/subselect_no_mat.result mysql-test/r/subselect_no_opts.result mysql-test/r/subselect_no_semijoin.result mysql-test/r/subselect_sj.result mysql-test/r/subselect_sj_jcl6.result mysql-test/t/ps.test mysql-test/t/subselect.test mysql-test/t/subselect3.test sql/opt_subselect.cc === modified file 'mysql-test/include/mix1.inc' --- a/mysql-test/include/mix1.inc 2009-09-15 06:08:54 +0000 +++ b/mysql-test/include/mix1.inc 2010-03-11 21:43:31 +0000 @@ -1177,8 +1177,11 @@ DROP TABLE t1; create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; --echo End of 5.0 tests === modified file 'mysql-test/r/index_merge_myisam.result' --- a/mysql-test/r/index_merge_myisam.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/index_merge_myisam.result 2010-03-11 21:43:31 +0000 @@ -1419,19 +1419,19 @@ drop table t1; # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge=off,index_merge_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='index_merge_union=on'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,index_merge_sort_union=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=off,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=4; ERROR 42000: Variable 'optimizer_switch' can't be set to the value of '4' set optimizer_switch=NULL; @@ -1458,21 +1458,21 @@ set optimizer_switch=default; set optimizer_switch='index_merge=off,index_merge_union=off,default'; select @@optimizer_switch; @@optimizer_switch -index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=off,index_merge_union=off,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set @@global.optimizer_switch=default; select @@global.optimizer_switch; @@global.optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on # # Check index_merge's @@optimizer_switch flags # select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, c int, filler char(100), @@ -1582,5 +1582,5 @@ id select_type table type possible_keys set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on drop table t0, t1; === modified file 'mysql-test/r/innodb_mysql.result' --- a/mysql-test/r/innodb_mysql.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mysql.result 2010-03-11 21:43:31 +0000 @@ -1425,12 +1425,15 @@ DROP TABLE t1; # create table t1 (a bit(1) not null,b int) engine=myisam; create table t2 (c int) engine=innodb; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch='partial_match_rowid_merge=off,partial_match_table_scan=off'; explain select b from t1 where a not in (select b from t1,t2 group by a) group by a; id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables 2 DEPENDENT SUBQUERY t1 system NULL NULL NULL NULL 0 const row not found 2 DEPENDENT SUBQUERY t2 ALL NULL NULL NULL NULL 1 +set optimizer_switch=@save_optimizer_switch; DROP TABLE t1,t2; End of 5.0 tests CREATE TABLE `t2` ( === modified file 'mysql-test/r/myisam_mrr.result' --- a/mysql-test/r/myisam_mrr.result 2010-01-17 14:51:10 +0000 +++ b/mysql-test/r/myisam_mrr.result 2010-03-11 21:43:31 +0000 @@ -394,7 +394,7 @@ drop table t0, t1; # - engine_condition_pushdown does not affect ICP select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (a int, b int, key(a)); === modified file 'mysql-test/r/ps.result' --- a/mysql-test/r/ps.result 2009-05-27 15:19:44 +0000 +++ b/mysql-test/r/ps.result 2010-03-11 21:43:31 +0000 @@ -149,6 +149,8 @@ c29 longblob, c30 longtext, c31 enum('on c32 set('monday', 'tuesday', 'wednesday') ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -177,6 +179,7 @@ id select_type table type possible_keys 2 DEPENDENT SUBQUERY NULL NULL NULL NULL NULL NULL NULL Impossible WHERE noticed after reading const tables deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; set @arg00=1; prepare stmt1 from ' create table t1 (m int) as select 1 as m ' ; execute stmt1 ; === modified file 'mysql-test/r/subselect.result' --- a/mysql-test/r/subselect.result 2010-02-17 21:59:41 +0000 +++ b/mysql-test/r/subselect.result 2010-03-11 21:43:31 +0000 @@ -1,4 +1,6 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4803,4 +4805,5 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. === modified file 'mysql-test/r/subselect3.result' --- a/mysql-test/r/subselect3.result 2010-02-17 10:05:27 +0000 +++ b/mysql-test/r/subselect3.result 2010-03-11 21:43:31 +0000 @@ -63,12 +63,15 @@ Handler_read_rnd_next 11 select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -692,6 +695,8 @@ a MAX(b) test 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -759,6 +764,7 @@ id select_type table type possible_keys 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -960,7 +966,7 @@ i1 i2 # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -977,7 +983,7 @@ i1 i2 # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect3_jcl6.result' --- a/mysql-test/r/subselect3_jcl6.result 2010-02-17 10:47:55 +0000 +++ b/mysql-test/r/subselect3_jcl6.result 2010-03-11 21:43:31 +0000 @@ -67,12 +67,15 @@ Handler_read_rnd_next 11 select ' ^ This must show 11' Z; Z ^ This must show 11 +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; id select_type table type possible_keys key key_len ref rows filtered Extra 1 PRIMARY t3 ALL NULL NULL NULL NULL 2 100.00 2 DEPENDENT SUBQUERY t1 ALL NULL NULL NULL NULL 6 100.00 Using where; Using temporary; Using filesort Warnings: Note 1003 select <in_optimizer>(`test`.`t3`.`a`,<exists>(select max(`test`.`t1`.`ie`) AS `max(ie)` from `test`.`t1` where (`test`.`t1`.`oref` = 4) group by `test`.`t1`.`grp` having trigcond((<cache>(`test`.`t3`.`a`) = <ref_null_helper>(max(`test`.`t1`.`ie`)))))) AS `a in (select max(ie) from t1 where oref=4 group by grp)` from `test`.`t3` +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; create table t1 (a int, oref int, key(a)); insert into t1 values @@ -696,6 +699,8 @@ a MAX(b) test 2 3 h 3 4 i DROP TABLE t1, t2; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; CREATE TABLE t1 (a int); CREATE TABLE t2 (b int, PRIMARY KEY(b)); INSERT INTO t1 VALUES (1), (NULL), (4); @@ -763,6 +768,7 @@ id select_type table type possible_keys 1 PRIMARY t1 ALL NULL NULL NULL NULL 4 Using where 2 DEPENDENT SUBQUERY t2 unique_subquery PRIMARY PRIMARY 4 func 1 Using index; Using where DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; CREATE TABLE t1 (a INT); INSERT INTO t1 VALUES(1); CREATE TABLE t2 (placeholder CHAR(11)); @@ -964,7 +970,7 @@ i1 i2 # Baseline: SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 17 +Handler_read_rnd_next 18 INSERT INTO t1 VALUES (NULL, NULL); FLUSH STATUS; @@ -981,7 +987,7 @@ i1 i2 # (read record from t1, but do not read from t2) SHOW STATUS LIKE '%Handler_read_rnd_next'; Variable_name Value -Handler_read_rnd_next 18 +Handler_read_rnd_next 19 DROP TABLE t1,t2; End of 5.1 tests CREATE TABLE t1 ( === modified file 'mysql-test/r/subselect_no_mat.result' --- a/mysql-test/r/subselect_no_mat.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_mat.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_opts.result' --- a/mysql-test/r/subselect_no_opts.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_opts.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='materialization=off,semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_no_semijoin.result' --- a/mysql-test/r/subselect_no_semijoin.result 2010-02-21 07:33:54 +0000 +++ b/mysql-test/r/subselect_no_semijoin.result 2010-03-11 21:43:31 +0000 @@ -1,8 +1,10 @@ show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='semijoin=off'; drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; select (select 2); (select 2) 2 @@ -4807,8 +4809,9 @@ SELECT 1 FROM t1 GROUP BY 1 1 DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; End of 5.1 tests. set optimizer_switch=default; show variables like 'optimizer_switch'; Variable_name Value -optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +optimizer_switch index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on === modified file 'mysql-test/r/subselect_sj.result' --- a/mysql-test/r/subselect_sj.result 2010-02-24 11:33:42 +0000 +++ b/mysql-test/r/subselect_sj.result 2010-03-11 21:43:31 +0000 @@ -202,39 +202,39 @@ BUG#37120 optimizer_switch allowable val select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/r/subselect_sj_jcl6.result' --- a/mysql-test/r/subselect_sj_jcl6.result 2010-03-07 15:41:45 +0000 +++ b/mysql-test/r/subselect_sj_jcl6.result 2010-03-11 21:43:31 +0000 @@ -206,39 +206,39 @@ BUG#37120 optimizer_switch allowable val select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,semijoin=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=on,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,semijoin=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=on,semijoin=off,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch='default,materialization=off,loosescan=off'; select @@optimizer_switch; @@optimizer_switch -index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on +index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_condition_pushdown=on,firstmatch=on,loosescan=off,materialization=off,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on set optimizer_switch=default; drop table t0, t1, t2; drop table t10, t11, t12; === modified file 'mysql-test/t/ps.test' --- a/mysql-test/t/ps.test 2009-05-27 15:19:44 +0000 +++ b/mysql-test/t/ps.test 2010-03-11 21:43:31 +0000 @@ -163,6 +163,9 @@ create table t1 ) engine = MYISAM ; create table t2 like t1; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + set @stmt= ' explain SELECT (SELECT SUM(c1 + c12 + 0.0) FROM t2 where (t1.c2 - 0e-3) = t2.c2 GROUP BY t1.c15 LIMIT 1) as scalar_s, exists (select 1.0e+0 from t2 where t2.c3 * 9.0000000000 = t1.c4) as exists_s, c5 * 4 in (select c6 + 0.3e+1 from t2) as in_s, (c7 - 4, c8 - 4) in (select c9 + 4.0, c10 + 40e-1 from t2) as in_row_s FROM t1, (select c25 x, c32 y from t2) tt WHERE x * 1 = c25 ' ; prepare stmt1 from @stmt ; execute stmt1 ; @@ -171,6 +174,8 @@ explain SELECT (SELECT SUM(c1 + c12 + 0. deallocate prepare stmt1; drop tables t1,t2; +set @@optimizer_switch=@save_optimizer_switch; + # # parameters from variables (for field creation) # === modified file 'mysql-test/t/subselect.test' --- a/mysql-test/t/subselect.test 2010-01-17 20:52:20 +0000 +++ b/mysql-test/t/subselect.test 2010-03-11 21:43:31 +0000 @@ -11,6 +11,9 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t11,t12; --enable_warnings +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + select (select 2); explain extended select (select 2); SELECT (SELECT 1) UNION SELECT (SELECT 2); @@ -4061,4 +4064,6 @@ SELECT 1 FROM t1 GROUP BY (SELECT LAST_INSERT_ID() FROM t1 ORDER BY MIN(a) ASC LIMIT 1); DROP TABLE t1; +set @@optimizer_switch=@save_optimizer_switch; + --echo End of 5.1 tests. === modified file 'mysql-test/t/subselect3.test' --- a/mysql-test/t/subselect3.test 2010-01-17 14:51:10 +0000 +++ b/mysql-test/t/subselect3.test 2010-03-11 21:43:31 +0000 @@ -59,9 +59,13 @@ select a in (select max(ie) from t1 wher show status like 'Handler_read_rnd_next'; select ' ^ This must show 11' Z; +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; + # This must show trigcond: explain extended select a in (select max(ie) from t1 where oref=4 group by grp) from t3; +set @@optimizer_switch=@save_optimizer_switch; drop table t1, t2, t3; # @@ -529,6 +533,9 @@ SELECT a, MAX(b), DROP TABLE t1, t2; +# The next three test cases must be executed with the IN=>EXISTS strategy +set @save_optimizer_switch=@@optimizer_switch; +set @@optimizer_switch="partial_match_rowid_merge=off,partial_match_table_scan=off"; # # Bug #27870: crash of an equijoin query with WHERE condition containing @@ -588,6 +595,8 @@ EXPLAIN SELECT a FROM t1 WHERE a NOT IN DROP TABLE t1, t2; +set @@optimizer_switch=@save_optimizer_switch; + # # Bug #34763: item_subselect.cc:1235:Item_in_subselect::row_value_transformer: # Assertion failed, unexpected error message: === modified file 'sql/opt_subselect.cc' --- a/sql/opt_subselect.cc 2010-03-09 10:36:15 +0000 +++ b/sql/opt_subselect.cc 2010-03-11 21:43:31 +0000 @@ -187,7 +187,11 @@ int check_and_do_in_subquery_rewrites(JO does not call setup_subquery_materialization(). We could make SELECT ... FROM DUAL call that function but that doesn't seem to be the case that is worth handling. - 4. Subquery is non-correlated + 4. Either the subquery predicate is a top-level predicate, or at + least one partial match strategy is enabled. If no partial match + strategy is enabled, then materialization cannot be used for + non-top-level queries because it cannot handle NULLs correctly. + 5. Subquery is non-correlated TODO: This is an overly restrictive condition. It can be extended to: (Subquery is non-correlated || @@ -195,13 +199,13 @@ int check_and_do_in_subquery_rewrites(JO (Subquery is correlated to the immediate outer query && Subquery !contains {GROUP BY, ORDER BY [LIMIT], aggregate functions}) && subquery predicate is not under "NOT IN")) - 5. No execution method was already chosen (by a prepared statement). + 6. No execution method was already chosen (by a prepared statement). (*) The subquery must be part of a SELECT statement. The current condition also excludes multi-table update statements. - We have to determine whether we will perform subquery materialization - before calling the IN=>EXISTS transformation, so that we know whether to + Determine whether we will perform subquery materialization before + calling the IN=>EXISTS transformation, so that we know whether to perform the whole transformation or only that part of it which wraps Item_in_subselect in an Item_in_optimizer. */ @@ -211,11 +215,14 @@ int check_and_do_in_subquery_rewrites(JO select_lex->master_unit()->first_select()->leaf_tables && // 3 thd->lex->sql_command == SQLCOM_SELECT && // * select_lex->outer_select()->leaf_tables && // 3A - subquery_types_allow_materialization(in_subs)) + subquery_types_allow_materialization(in_subs) && + // psergey-todo: duplicated_subselect_card_check: where it's done? + (in_subs->is_top_level_item() || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_ROWID_MERGE) || + optimizer_flag(thd, OPTIMIZER_SWITCH_PARTIAL_MATCH_TABLE_SCAN)) &&//4 + !in_subs->is_correlated && // 5 + in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 6 { - // psergey-todo: duplicated_subselect_card_check: where it's done? - if (!in_subs->is_correlated && // 4 - in_subs->exec_method == Item_in_subselect::NOT_TRANSFORMED) // 5 in_subs->exec_method= Item_in_subselect::MATERIALIZATION; }

1 0

Re: [Maria-developers] options for CREATE TABLE (MWL#43)
by Oleksandr Byelkin 11 Mar '10

11 Mar '10

Hi! Some you request contradicts with Mony's ones so I think it should be discussed somehow. 11 марта 2010, в 12:30, Sergei Golubchik написал(а): > Hi, Sanja! > > Here's the review, below: > > Summary: > > 1. please, store options together with the objects they describe, not > separately. I tried to do so in my very first implementation and IMHO it was my mistake. The same related to way of storing elements - it should be the same for parsing and reading. During parsing we one one structures and classes, for storing in TABLE/TABLE_SHARE others. Moving data between them (taking into account different memroot where they are allocated) was quite tricky. I can make it again if you both think that it is really important. > 2. Unknown option should be an error by default. OK. The only problem is that it is contradict to Monty requirements. Our initial decision was issue error if option was added explicitly. The only problem is that it is very difficult to implement - we write options to .frm first then read them and pass to engine. I have no idea how to pass this information via/over frm. > 3. use something my_getopt-like as we discussed, don't force every > engine to parse its options I can add such function for users to use, but it will be thier choice use it or do not, is it OK? > 4. make options immutable to avoid copying them in ::clone I do not know way to do it if they should be allocated in different mem_roots. > 5. don't check for changed options in alter table with your > check_if_incompatible_data. let the engine do that. This and 8 require big changes engine and ALTER TEBLE. Monty's requirement was do not touch current code. I would be glad if you discuss it and make some non contradicting requirement. > 6. parser: use ident, not IDENT_sys OK > 7. parser: make the equal sign optional I have some doubts that it is doable DATA DIRECTORY TEST VALUE ... Does it mean: DATA = DIRECTORY TEST = VALUE ... or DATA DIRECTORY = TEST VALUE ... ? - error (ALTER TABLE uses create_table_options_space_separated list of options) Other problem is should we store old options in new way, old way, both. (I think in this case both). > 8. few existing options, like row_format, insert_method, checksum, > delay_key_write, key_block_size, min_rows/max_rows, avg_row_length, > tablespace, connection, pack_keys could be moved into storage > engines > out of the parser. See above. > 9. make sure your code works (and tested) with table options specified > per partition/subpartition OK. > 10. misc details, like using 'changed' or unnecessary complex encoding > of options in the frm file, see below. > >> === added file 'mysql-test/r/create_options.result' >> --- mysql-test/r/create_options.result 1970-01-01 00:00:00 +0000 >> +++ mysql-test/r/create_options.result 2010-03-04 20:46:55 +0000 >> @@ -0,0 +1,197 @@ >> +drop table if exists t1; >> +create table t1 (a int fkey1=v1, key akey (a) kkey1=v1) tkey1=1v1 >> TKEY1=NULL tkey1=1v2 tkey2=2v1 tkey3=3v1; >> +Warnings: >> +Warning 1650 Unused option 'tkey1'='1v2' >> +Warning 1650 Unused option 'tkey2'='2v1' >> +Warning 1650 Unused option 'tkey3'='3v1' >> +Warning 1651 Unused option 'fkey1'='v1' of field 'a' >> +Warning 1652 Unused option 'kkey1'='v1' of key 'akey' > > 1. Better "unknown" or "unsupported" e.g. > > Unknown option 'tkey1' > Unsupported option 'fkey1' specified for field 'a' > Invalid option 'kkey1' used for key 'akey' > > no, "invalid" is bad here, scratch that ok > > 2. why there's no warning for TKEY1=NULL ? Because it means remove option. > >> +drop table t1; >> +create table t1 (a int fkey1=v1, key akey (a) kkey1=v1) tkey1=1v1 >> tkey1=1v2 TKEY1=NULL tkey1=1v1 tkey1=1v2 tkey2=2v1 tkey3=3v1; > > I don't understand how this is different from the first test > (and many of the tests bellow), > could you please add short one-line comments to the .test file keys are in different order. > explaining what you test in each statement ? OK > > also, a thought about "warning vs errors": > making warnings for typos and unknown options is one of the most > disliked features in MySQL - judging from the number of bugreports > (bug reports about USE HASH/BTREE, mind you - only a couple of places > where MySQL is promiscuous like that, guess what will happen when your > patch will take it to a whole new level!). > > moving engines, and so on, I know - but most users don't care. > > STRICT mode is too strict here, I think, it adds too much strictness > everywhere. What about adding a special mode that's only "strict" in > create > table (and alter table - user specified part) ? That should be ON by > default > (or, rather, a negative mode should be OFF by default). > > In other words - I want the patch to be optimized (performance, and > user > experience) for the common case, not to boundary cases. And the common > case, I believe, is the one when a user does not change engines all > the > time. We support the boundary case, yes, but optimize for the common > one. You remember that I also was for errors, but MOnty still want warnings. Also there is problem in implementation of the way we agreed on (see abopve about ALTER TABLE). >> +Warnings: >> +Warning 1650 Unused option 'tkey1'='1v2' >> +Warning 1650 Unused option 'tkey2'='2v1' >> +Warning 1650 Unused option 'tkey3'='3v1' >> +Warning 1651 Unused option 'fkey1'='v1' of field 'a' >> +Warning 1652 Unused option 'kkey1'='v1' of key 'akey' >> +drop table t1; >> +create table t1 (a int fkey1=v1, key akey (a) kkey1=v1) tkey1=1v1 >> tkey1=1v2 TKEY1='NULL' tkey2=2v1 tkey3=3v1; > ... >> === added file 'mysql-test/t/create_options_example.test' >> --- mysql-test/t/create_options_example.test 1970-01-01 00:00:00 >> +0000 >> +++ mysql-test/t/create_options_example.test 2010-03-04 20:46:55 >> +0000 >> @@ -0,0 +1,16 @@ >> +--source include/have_example_plugin.inc >> + >> +--disable_warnings >> +drop table if exists t1; >> +--enable_warnings >> + >> +#All vaues with warnings > > this should go into plugin.test or exampledb.test Why not separate test? >> +create table t1 (a int ttt=xxx E=1, key akey (a) kkk=xxx ) E=1 >> ttt=xxx ttt=yyy TTT=DEFAULT mmm=CCC zzz=MMM; >> + >> +drop table t1; >> + >> +# E=1 accepted by engine >> +create table t1 (a int ttt=xxx E=1) ENGINE=EXAMPLE E=1 ttt=xxx >> ttt=yyy TTT=DEFAULT mmm=CCC zzz=MMM; >> + >> +drop table t1; >> + >> === modified file 'sql/Makefile.am' >> --- sql/Makefile.am 2010-03-03 14:44:14 +0000 >> +++ sql/Makefile.am 2010-03-04 20:46:55 +0000 >> @@ -124,7 +124,7 @@ mysqld_SOURCES = sql_lex.cc sql_handler. >> sql_plugin.cc sql_binlog.cc \ >> sql_builtin.cc sql_tablespace.cc >> partition_info.cc \ >> sql_servers.cc event_parse_data.cc \ >> - opt_table_elimination.cc >> + opt_table_elimination.cc >> sql_create_options.cc > > please make sure that 'make distcheck' works after your changes OK >> >> nodist_mysqld_SOURCES = mini_client_errors.c pack.c client.c >> my_time.c my_user.c >> >> === modified file 'storage/example/ha_example.cc' >> --- storage/example/ha_example.cc 2010-03-03 14:44:14 +0000 >> +++ storage/example/ha_example.cc 2010-03-04 20:46:55 +0000 >> @@ -836,11 +836,43 @@ ha_rows ha_example::records_in_range(uin >> int ha_example::create(const char *name, TABLE *table_arg, >> HA_CREATE_INFO *create_info) >> { >> + CREATE_OPTION *opt; >> DBUG_ENTER("ha_example::create"); >> /* >> This is not implemented but we want someone to be able to see >> that it >> works. >> */ >> + /* Example of checking parameters for table*/ >> + if (!create_info->create_table_options) >> + DBUG_RETURN(0); >> + for (opt= create_info->create_table_options->table_opt.first; >> + opt; >> + opt= opt->next) >> + { >> + /* check for legal options and its legal values */ >> + if (opt->key.length == 1 && >> + (opt->key.str[0] == 'e' || opt->key.str[0] == 'E') && >> + opt->val.length == 1 && >> + opt->val.str[0] == '1') >> + opt->used= 1; /* tell MariaDB that we used the only legal >> parameter */ >> + } >> + /* Example of checking parameters for fields*/ >> + for (Field **field= table_arg->s->field; *field; field++) >> + { >> + if ((*field)->create_options.first) >> + { >> + for (opt= (*field)->create_options.first; opt; opt= opt->next) >> + { >> + /* check for legal options and its legal values */ >> + if (opt->key.length == 1 && >> + (opt->key.str[0] == 'e' || opt->key.str[0] == 'E') && >> + opt->val.length == 1 && >> + opt->val.str[0] == '1') >> + opt->used= 1; /* tell MariaDB that we used the only >> legal parameter */ >> + } >> + } >> + } > > No, that's way too complex and too much code. > *every* engine will need to do that, which means - it should be done > in the > server for all engines. Why you didn't use my_getopt as we originally > discussed ? OK (see above) >> + >> DBUG_RETURN(0); >> } >> >> === added file 'sql/sql_create_options.h' >> --- sql/sql_create_options.h 1970-01-01 00:00:00 +0000 >> +++ sql/sql_create_options.h 2010-03-04 20:46:55 +0000 >> @@ -0,0 +1,102 @@ >> + >> +#ifndef _SQL_CREATE_OPTIONS_H >> +#define _SQL_CREATE_OPTIONS_H >> + >> + >> +/* types of cretate options records on disk, also it is length of >> extra data */ > > 1. typo: create > 2. I know what does "length of extra data" mean, but the comment does > not help to understand it. I just forgot to change comment after changing parameter (monty wanted 2 bytes for key number also just tu be able to increase number of keys) >> +typedef enum enum_create_options_type { >> + CREATE_OPTION_TABLE= 0, >> + CREATE_OPTION_KEY= 1, >> + CREATE_OPTION_FIELD= 2 >> +} CREATE_OPTION_TYPES; >> + >> +typedef struct st_create_option { >> + /* pointer to the next option or NULL */ >> + struct st_create_option *next; >> + /* pointer to Field or KEY or NULL */ >> + void *owner; > > 1. better to use union { Field *, KEY *} > 2. even better - use 'const char* name' as you don't need anything > else > from your fields/keys here > 3. even better remove this 'owner' at all, you don't need it - see > below, > if you iterate the list of fields and keys you always know > what field/key the option belongs to. OK >> + /* key and value of the option (\0 terminated)*/ >> + LEX_STRING key, val; >> + /* used to issue warnings about unused options */ >> + my_bool used; >> +} CREATE_OPTION; >> + >> +struct st_table_options; >> + >> + >> +class st_create_option_list { > > why did you need to create your own list implementation instead of > using > either one that MySQL already has ? > (hint: LIST, I_List, List, or even dynamic array) > > and - > why do you need any list at all, if you store options in Fields and > KEYs > and simply can use the existing lists of fields and keys ? Historically. But yes now it would be better to use LIST (if it allow to insert item at the end)... >> +public: >> + /** >> + pointer on the first list element >> + */ >> + CREATE_OPTION *first; >> + /** >> + pointer on last list '.next' or beginning of the list in case >> of empty list >> + >> + @note: >> + If it is NULL then it is just sign of array of list end >> + */ >> +private: >> + CREATE_OPTION **last; >> +public: >> + void empty() {first= NULL; last= &first;} >> + st_create_option_list() {empty();} >> + st_create_option_list(const st_create_option_list &o) >> + { >> + if ((first= o.first)) >> + last= o.last; >> + else >> + last= &first; >> + } >> + my_bool last_opt() { return last == NULL; } >> + friend my_bool create_option_add(st_create_option_list *options, >> + MEM_ROOT *root, >> + const LEX_STRING *str_key, >> + const LEX_STRING *str_val, >> + my_bool *changed); >> + friend st_create_option_list >> *create_create_options_array(MEM_ROOT *root, >> + uint n); >> + friend my_bool create_options_read(const uchar *buff, uint length, >> + MEM_ROOT *root, >> + st_table_options *opt); >> + friend my_bool create_options_clone(MEM_ROOT *root, >> + st_create_option_list *opts); >> +}; >> +typedef class st_create_option_list CREATE_OPTION_LIST; >> + >> + >> +typedef struct st_table_options { >> + CREATE_OPTION_LIST table_opt; /* table options list */ >> + CREATE_OPTION_LIST *field_opt; /* fields options array */ >> + CREATE_OPTION_LIST *key_opt; /* keys options array */ >> +} TABLE_OPTIONS; >> + >> +CREATE_OPTION_LIST *create_create_options_array(MEM_ROOT *root, >> uint n); >> +TABLE_OPTIONS *create_create_options(MEM_ROOT *root, uint fields, >> uint keys); >> + >> +my_bool create_options_read(const uchar *buff, uint length, >> MEM_ROOT *root, >> + TABLE_OPTIONS *opt); >> + >> +my_bool create_option_add(CREATE_OPTION_LIST *options, MEM_ROOT >> *root, >> + const LEX_STRING *k, const LEX_STRING *v, >> + my_bool *chanes); >> + >> +ulong create_options_length(TABLE_OPTIONS *opt); >> + >> +void create_options_store(uchar *buff, TABLE_OPTIONS *opt); >> + >> +void create_options_check_unused(THD *thd, >> + TABLE_OPTIONS *options); >> + >> +struct st_table_share; >> +void create_options_binding(struct st_table_share *share); >> + >> +my_bool create_options_clone(MEM_ROOT *root, CREATE_OPTION_LIST >> *opt); >> + >> +CREATE_OPTION_LIST *create_table_list_merge(CREATE_OPTION_LIST >> *source, >> + CREATE_OPTION_LIST >> *changes, >> + MEM_ROOT *root, >> + my_bool *changed); >> +my_bool is_equal_create_options(CREATE_OPTION *opt1, CREATE_OPTION >> *opt2); >> + >> +#endif >> === modified file 'sql/table.h' >> --- sql/table.h 2010-02-12 08:47:31 +0000 >> +++ sql/table.h 2010-03-04 20:46:55 +0000 >> @@ -340,6 +340,7 @@ typedef struct st_table_share >> #ifdef NOT_YET >> struct st_table *open_tables; /* link to open tables */ >> #endif >> + TABLE_OPTIONS *create_table_options; /* text options for table */ > > do you need TABLE_OPTIONS - I mean, table, field, and key options - > here ? TABLE_SHARE has an array of KEYs and KEYs store options > internally (in KEY::create_options). And exactly the same > applies to Fields. I described it in the beginning. >> >> /* The following is copied to each TABLE on OPEN */ >> Field **field; >> === modified file 'sql/structs.h' >> --- sql/structs.h 2010-02-01 06:14:12 +0000 >> +++ sql/structs.h 2010-03-04 20:46:55 +0000 >> @@ -101,6 +101,8 @@ typedef struct st_key { >> int bdb_return_if_eq; >> } handler; >> struct st_table *table; >> + /** reference to the list of options or NULL */ >> + CREATE_OPTION_LIST create_options; > > eh, strictly speaking 'create_options' is not a pointer and it > cannot be NULL. > And it is not a reference in the C++ sense either. > > you could've simply said "list of options" Not fixed comment, sorry. >> } KEY; >> >> >> === modified file 'sql/handler.h' >> --- sql/handler.h 2010-02-01 06:14:12 +0000 >> +++ sql/handler.h 2010-03-04 20:46:55 +0000 >> @@ -919,6 +919,12 @@ typedef struct st_ha_create_information >> LEX_STRING connect_string; >> const char *password, *tablespace; >> LEX_STRING comment; >> + TABLE_OPTIONS create_table_options_orig; >> + /** >> + Originally create_table_options points on above field, but >> during ALTER >> + TABLE of the options it points on new built parameters >> + */ >> + TABLE_OPTIONS *create_table_options; > > after reading the patch I still don't understand why do you need > create_table_options_orig For avoiding allocating it for normal table, it will be changed in ALTER TABLE process. >> const char *data_file_name, *index_file_name; >> const char *alias; >> ulonglong max_rows,min_rows; >> === modified file 'sql/sql_class.cc' >> --- sql/sql_class.cc 2010-02-01 06:14:12 +0000 >> +++ sql/sql_class.cc 2010-03-04 20:46:55 +0000 >> @@ -109,6 +109,8 @@ Key::Key(const Key &rhs, MEM_ROOT *mem_r >> generated(rhs.generated) >> { >> list_copy_and_replace_each_value(columns, mem_root); >> + create_options= rhs.create_options; >> + create_options_clone(mem_root, &create_options); > > in create_options_clone() you don't need to clone everything, > this constructor only copies elements that can change during > execution, > for example field and key names don't change and don't need to be > copied. And options don't change either, only their "used" property > is. > but it would be best if you would get rid of it and make options > completely > immutable. There was problems like pointer on freed memory which gone after this I suspect different mem_roots. >> } >> >> /** >> === modified file 'sql/field.h' >> --- sql/field.h 2010-02-01 06:14:12 +0000 >> +++ sql/field.h 2010-03-04 20:46:55 +0000 >> @@ -137,6 +137,8 @@ class Field >> struct st_table *table; // Pointer for table >> struct st_table *orig_table; // Pointer to original table >> const char **table_name, *field_name; >> + /** reference to the list of options or NULL */ > > this is neither a reference nor it can be NULL old comment >> + CREATE_OPTION_LIST create_options; >> LEX_STRING comment; >> /* Field is part of the following keys */ >> key_map key_start, part_of_key, part_of_key_not_clustered; >> === modified file 'sql/field.cc' >> --- sql/field.cc 2010-02-01 06:14:12 +0000 >> +++ sql/field.cc 2010-03-04 20:46:55 +0000 >> @@ -10220,6 +10225,7 @@ Create_field::Create_field(Field *old_fi >> decimals= old_field->decimals(); >> vcol_info= old_field->vcol_info; >> stored_in_db= old_field->stored_in_db; >> + create_options= old_field->create_options; > > explain in a comment please why you don't need to copy the data > here, and can simply assign pointers Because copy constructor makes correct list assignment, is it correct comment? > >> >> /* Fix if the original table had 4 byte pointer blobs */ >> if (flags & BLOB_FLAG) >> === modified file 'sql/sql_show.cc' >> --- sql/sql_show.cc 2010-02-01 06:14:12 +0000 >> +++ sql/sql_show.cc 2010-03-04 20:46:55 +0000 >> @@ -1356,6 +1376,8 @@ int store_create_info(THD *thd, TABLE_LI >> packet->append(STRING_WITH_LEN(" COMMENT ")); >> append_unescaped(packet, field->comment.str, field- >> >comment.length); >> } >> + if (field->create_options.first) > > you don't need an if() here and below, append_create_options() > can handle the case of create_options.first == 0 OK (but i will need change list implementation in any case) > >> + append_create_options(thd, packet, field- >> >create_options.first); >> } >> >> key_info= table->key_info; >> @@ -1586,6 +1610,11 @@ int store_create_info(THD *thd, TABLE_LI >> packet->append(STRING_WITH_LEN(" CONNECTION=")); >> append_unescaped(packet, share->connect_string.str, share- >> >connect_string.length); >> } >> + /* create_table_options can be NULL for temporary tables */ >> + if (share->create_table_options && > > why TABLE_SHARE::create_table_options is a pointer to something > allocared > on TABLE_SHARE::mem_root ? In Field and KEY it's simply > a structure - part of the Field/KEY class, why not the same here ? Most time it is pointer to create_table_options_orig. It is not the same here because of ALTER TABLE and the way how it plays with TABLE_SHARE. > >> + share->create_table_options->table_opt.first) >> + append_create_options(thd, packet, >> + share->create_table_options- >> >table_opt.first); >> append_directory(thd, packet, "DATA", >> create_info.data_file_name); >> append_directory(thd, packet, "INDEX", >> create_info.index_file_name); >> } >> === modified file 'sql/sql_table.cc' >> --- sql/sql_table.cc 2010-02-12 08:47:31 +0000 >> +++ sql/sql_table.cc 2010-03-04 20:46:55 +0000 >> @@ -5789,6 +5791,15 @@ compare_tables(TABLE *table, >> DBUG_RETURN(0); >> } >> >> + if (!is_equal_create_options(tmp_new_field- >> >create_options.first, >> + field->create_options.first)) >> + { > > I am not sure this should be checked on MySQL level, we don't know the > semantics of options. I'd say this check belong to > handler::check_if_incompatible_data() and should be implemented in the > storage engine internally. Monty even requested me to recreate .frm even if case of KEY was chenged (which is clear do not chengr semantic) - i.e. any change == rewriting .frm. So your requests contradict here it should be discussed (I do not see sens nor harm in such rewriting policy) >> + DBUG_PRINT("info", ("Options difference in field '%s'", >> + new_field->field_name)); >> + *need_copy_table= ALTER_TABLE_DATA_CHANGED; >> + DBUG_RETURN(0); >> + } >> + >> /* Don't pack rows in old tables if the user has requested >> this. */ >> if (create_info->row_type == ROW_TYPE_DYNAMIC || >> (tmp_new_field->flags & BLOB_FLAG) || >> @@ -6112,6 +6125,41 @@ mysql_prepare_alter_table(THD *thd, TABL >> } >> restore_record(table, s->default_values); // Empty record for >> DEFAULT >> >> + if (create_info->create_table_options_orig.table_opt.first) >> + { >> + CREATE_OPTION_LIST *res; >> + my_bool changed= FALSE; >> + if (!table->s->create_table_options && >> + !(table->s->create_table_options= >> + create_create_options(&table->s->mem_root, >> + table->s->fields, table->s->keys))) >> + goto err; >> + >> + if (!(res= >> + create_table_list_merge(&table->s->create_table_options- >> >table_opt, >> + &create_info-> >> + >> create_table_options_orig.table_opt, >> + thd->mem_root, >> + &changed))) >> + goto err; >> + DBUG_ASSERT(res->first); >> + create_info->create_table_options_orig.table_opt= *res; >> + >> + if (changed) >> + alter_info->change_level= ALTER_TABLE_DATA_CHANGED; >> + else >> + { >> + alter_info->flags&= ~ALTER_CREATE_OPT; >> + DBUG_PRINT("info", ("Table options was not changed")); >> + } >> + } >> + else >> + if (table->s->create_table_options) >> + create_info->create_table_options_orig.table_opt= >> + table->s->create_table_options->table_opt; > > why don't you set ALTER_TABLE_DATA_CHANGED here ? it used as flag from parser only. > >> + else >> + create_info->create_table_options_orig.table_opt.empty(); >> + >> /* >> First collect all fields from table which isn't in drop_list >> */ >> === modified file 'sql/sql_yacc.yy' >> --- sql/sql_yacc.yy 2010-02-01 06:14:12 +0000 >> +++ sql/sql_yacc.yy 2010-03-04 20:46:55 +0000 >> @@ -4714,6 +4718,16 @@ create_table_option: >> Lex->create_info.used_fields|= >> HA_CREATE_USED_TRANSACTIONAL; >> Lex->create_info.transactional= $3; >> } >> + | IDENT_sys equal plugin_option_value > > 1. why IDENT_sys and not ident ? OK > 2. perhaps we should make the equal sign optional ? > first - that's backward compatible, > second - that would allow us to simplify the code quite a bit, > moving existing table and index options onto a new framework Answered above. >> + { >> + LEX *lex= Lex; >> + create_option_add(&(lex-> >> + create_info. >> + >> create_table_options_orig.table_opt), >> + YYTHD->mem_root, &$1, &$3, >> + NULL); >> + lex->alter_info.flags|= ALTER_CREATE_OPT; >> + } >> ; >> >> default_charset: >> @@ -13827,6 +13867,32 @@ uninstall: >> } >> ; >> >> +/ >> ************************************************************************** >> + >> + Create options >> + >> + >> **************************************************************************/ >> + >> +plugin_option_value: >> + DEFAULT >> + { >> + $$.str= NULL; /* We are going to remove the option */ >> + $$.length= 0; >> + } >> + | NULL_SYM > > I don't like this trick. > If you don't support NULLs, dont't allow users to specify them how it can be stored as parameter value? Such semantic prevent users of thinking that assigning NULL will make it really NULL not "NULL". >> + { >> + $$.str= NULL; /* We are going to remove the option */ >> + $$.length= 0; >> + } >> + | IDENT_sys { $$ = $1; } >> + | TEXT_STRING_sys { $$ = $1; } >> + | DECIMAL_NUM { $$ = $1; } >> + | FLOAT_NUM { $$ = $1; } >> + | NUM { $$ = $1; } >> + | LONG_NUM { $$ = $1; } >> + | HEX_NUM { $$ = $1; } > > looks like you forgot a semicolon here OK >> + >> + >> /** >> @} (end of group Parser) >> */ >> >> === added file 'sql/sql_create_options.cc' >> --- sql/sql_create_options.cc 1970-01-01 00:00:00 +0000 >> +++ sql/sql_create_options.cc 2010-03-04 20:46:55 +0000 >> @@ -0,0 +1,646 @@ >> + >> +#include "mysql_priv.h" >> + >> +/* Additional length of index for CREATE_OPTION_XXX types */ > > the comment is confusing. I could understand from the code what > create_options_len[] is for, but the comment did not help in the least "Length of additional data stored for every CREATE_OPTION_XXX types " Is it OK? > >> +static uint create_options_len[3]= {0, 2, 2}; >> + >> + >> +/** >> + Adds new option to this list >> + >> + @param options pointer to the list >> + @param root memroot to allocate option >> + @param str_key key >> + @param str_val value >> + @param changed pointer to variable to report changed data >> + >> + @retval TRUE error >> + @retval FALSE OK >> +*/ >> + >> +my_bool create_option_add(CREATE_OPTION_LIST *options, MEM_ROOT >> *root, >> + const LEX_STRING *str_key, >> + const LEX_STRING *str_val, >> + my_bool *changed) >> +{ >> + CREATE_OPTION *cur_option, **option; >> + char *key, *val; >> + my_bool not_used; >> + my_bool copy= FALSE; >> + my_bool replace= FALSE; >> + DBUG_ENTER("create_option_add"); >> + DBUG_PRINT("enter", ("key: '%s' value: '%s'", >> + str_key->str, str_val->str)); >> + if (changed) >> + copy= TRUE; >> + else >> + changed= &not_used; >> + >> + DBUG_ASSERT(options->first || >> + (!options->first && options->last == &options- >> >first)); >> + *changed= FALSE; > > Hmm, strange. From the way you use 'changed' I thought it should > accumulate > the results - I mean, it's one variable that is passed into > create_option_add() for all options. Apparently at the end it should > be > true if *any* of the options has changed. > > But then, why do you set it to false inside create_option_add() ? It was special case for call from ALTER TABLE and from parser. Only ALTER TABLE was interested in changes and so required copying parameters. >> + >> + /* try to find the option first */ >> + for (option= &(options->first); >> + *option && my_strcasecmp(system_charset_info, >> + str_key->str, (*option)->key.str); >> + option= &((*option)->next)) ; >> + if (str_val->str) >> + { >> + /* add / replace */ >> + if (*option) >> + { >> + /* replace */ >> + cur_option= *option; >> + if (!(*changed) && >> + (cur_option->val.length != str_val->length || >> + memcmp(cur_option->val.str, str_val->str, str_val- >> >length))) >> + { >> + *changed= TRUE; >> + } >> + replace= TRUE; >> + } >> + else >> + { >> + /* add */ >> + if (!(cur_option= (CREATE_OPTION *)alloc_root(root, >> + >> sizeof(CREATE_OPTION)))) >> + DBUG_RETURN(TRUE); >> + bzero(cur_option, sizeof(CREATE_OPTION)); >> + *(options->last)= cur_option; >> + options->last= &(cur_option->next); >> + *changed= TRUE; >> + } >> + if (changed || replace) >> + { >> + /* >> + In case of replace we use new key in case it differ only >> in case >> + like 'key' and 'KEY' >> + */ >> + if (!multi_alloc_root(root, &key, str_key->length + 1, >> + &val, str_val->length + 1, NULL)) >> + DBUG_RETURN(TRUE); >> + cur_option->key.str= >> + (char *)memcpy(key, str_key->str, >> + (cur_option->key.length= str_key->length)); >> + key[str_key->length]= '\0'; >> + cur_option->val.str= >> + (char *)memcpy(val, str_val->str, >> + (cur_option->val.length= str_val->length)); >> + val[str_val->length]= '\0'; >> + cur_option->used= FALSE; >> + cur_option->owner= NULL; >> + } >> + DBUG_ASSERT(options->first || >> + (!options->first && options->last == &options- >> >first)); >> + } >> + else >> + { >> + /* remove */ >> + if (*option) >> + { >> + if (options->last == &((*option)->next)) >> + options->last= option; /* we deleted last option */ >> + *option= (*option)->next; >> + *changed= TRUE; >> + DBUG_ASSERT(options->first || >> + (!options->first && options->last == &options- >> >first)); >> + } >> + } >> + DBUG_RETURN(FALSE); >> +} >> + >> + >> +/** >> + Creates empty fields/keys array for table create options structure >> + >> + @param root memroot where to allocate memory for this >> structure >> + @param n number of fields/keys >> + >> + @return pointer to array or NULL in case of error. >> +*/ >> + >> +CREATE_OPTION_LIST *create_create_options_array(MEM_ROOT *root, >> uint n) > > "create_create" is not a good name :( I did not found better but open for suggestion. > >> +{ >> + uint i; >> + DBUG_ENTER("create_create_options_array"); >> + DBUG_PRINT("enter", ("Number: %u", n)); >> + >> + CREATE_OPTION_LIST *res= >> + (CREATE_OPTION_LIST *) alloc_root(root, >> + sizeof(CREATE_OPTION_LIST) * (n >> + 1)); >> + bzero(res, sizeof(CREATE_OPTION_LIST) * (n + 1)); >> + if (!res) >> + DBUG_RETURN(NULL); >> + for (i= 0; i < n; i++) >> + res[i].last= &res[i].first; >> + /* We do not do above for res[n]. It is sign of array end */ >> + DBUG_RETURN(res); >> +} >> + >> + >> +/** >> + Reads options from this buffer >> + >> + @param buffer the buffer to read from >> + @param mem_root memroot for allocating >> + @param opt parametes to write to >> + >> + @retval TRUE Error >> + @retval FALSE OK >> +*/ >> + >> +my_bool create_options_read(const uchar *buff, uint length, >> MEM_ROOT *root, >> + TABLE_OPTIONS *opt) >> +{ >> + const uchar *buff_end= buff + length; >> + DBUG_ENTER("create_options_read"); >> + while (buff < buff_end) >> + { >> + CREATE_OPTION *option; >> + CREATE_OPTION_TYPES type; >> + uint index= 0; >> + >> + if (!(option= (CREATE_OPTION *) alloc_root(root, >> sizeof(CREATE_OPTION)))) >> + DBUG_RETURN(TRUE); >> + >> + DBUG_ASSERT(buff + 4 <= buff_end); >> + option->val.length= uint2korr(buff); >> + option->key.length= buff[2]; >> + option->next= NULL; >> + type= (CREATE_OPTION_TYPES)buff[3]; >> + buff+= 4; >> + switch (type) { >> + case CREATE_OPTION_FIELD: > > interesting encoding. so basically you support the case when field, > key, and table options are all written interleaved: > > <table option><key 1 option><field 5 option><table option><field 3 > option><key 4 option>... > > why the heck do you want to support it ? Could you propose other encoding taking into account that some fields, keys and tables do not have parameters and some has several ones? >> + index= uint2korr(buff); >> + buff+= 2; >> + *(opt->field_opt[index].last)= option; >> + opt->field_opt[index].last= &option->next; >> + break; >> + case CREATE_OPTION_KEY: >> + index= uint2korr(buff); >> + buff+= 2; >> + *(opt->key_opt[index].last)= option; >> + opt->key_opt[index].last= &option->next; >> + break; >> + case CREATE_OPTION_TABLE: >> + /* table */ >> + *(opt->table_opt.last)= option; >> + opt->table_opt.last= &option->next; >> + break; >> + default: >> + DBUG_ASSERT(0); >> + } >> + if (!(option->key.str= strmake_root(root, (const char*)buff, >> + option->key.length))) >> + DBUG_RETURN(TRUE); >> + buff+= option->key.length; >> + if (!(option->val.str= strmake_root(root, (const char*)buff, >> + option->val.length))) >> + DBUG_RETURN(TRUE); >> + buff+= option->val.length; >> + option->used= FALSE; >> + option->owner= NULL; >> + DBUG_PRINT("info", ("type: %u index: %u key: '%s' value: >> '%s'", >> + (uint) type, (uint) index, >> + option->key.str, option->val.str)); >> + } >> + DBUG_RETURN(FALSE); >> +} >> + >> +/** >> + Calculates length of saved image of the option lists >> + >> + @param opt list of options >> + @param extra_length type of the record > > eh, extra_length is not really a "type of the record", is it ? it was, but you are right it should be fixed. >> + >> + @return length >> +*/ >> + >> +static ulong create_options_list_length(CREATE_OPTION_LIST *opts, >> int extra_length) >> +{ >> + CREATE_OPTION *opt; >> + ulong res= 0; >> + DBUG_ENTER("create_options_list_length"); >> + for (opt= opts->first; opt != NULL; opt= opt->next) >> + { >> + DBUG_PRINT("info", ("key: '%s' value: '%s'", >> + (opt->key.str ? opt->key.str : "<NULL>"), >> + (opt->val.str ? opt->val.str : "<NULL>"))); >> + DBUG_ASSERT(opt->key.length); >> + /* >> + length of disk for every record: >> + 2 bytes - value length >> + 1 byte - key length >> + 1 byte - record type >> + 0/2 bytes - none/key number/field number >> + */ >> + res+= 2 + 1 + 1 + extra_length + opt->key.length + opt- >> >val.length; >> + } >> + DBUG_RETURN(res); >> +} >> + >> +/** >> + Calculates length of saved image of the all options of the table >> + >> + @param opts table of options >> + >> + @return length >> +*/ >> + >> +ulong create_options_length(TABLE_OPTIONS *opt) >> +{ >> + CREATE_OPTION_LIST *opts; >> + ulong res; >> + DBUG_ENTER("create_options_length"); >> + >> + res= >> + (opt->table_opt.first ? >> + create_options_list_length(&opt->table_opt, >> + >> create_options_len[CREATE_OPTION_TABLE]): >> + 0); >> + if (opt->field_opt) >> + { >> + for (opts= opt->field_opt; !opts->last_opt(); opts++) > > why wouldn't you simply iterate over an array of the fixed length - > you know how many fields and keys are there. And you wouldn't need > this "invalid list" array element at the end. To avoid knowing too much about other structures and classes. > even better - as I wrote above, keep options together with fields/ > keys only > and don't maintain a separate array of them. I explained what problems it brings if you think that it is vitally important I will make it. >> + res+= >> + create_options_list_length(opts, >> + >> create_options_len[CREATE_OPTION_FIELD]); >> + } >> + if (opt->key_opt) >> + { >> + for (opts= opt->key_opt; !opts->last_opt(); opts++) >> + res+= >> + create_options_list_length(opts, >> + >> create_options_len[CREATE_OPTION_KEY]); >> + } >> + DBUG_RETURN(res); >> +} > > > Regards, > Sergei

1 0

[Maria-developers] Rev 2734: Maria WL#61 in file:///Users/bell/maria/bzr/work-maria-5.2-engine/
by sanja＠askmonty.org 11 Mar '10

11 Mar '10

At file:///Users/bell/maria/bzr/work-maria-5.2-engine/ ------------------------------------------------------------ revno: 2734 revision-id: sanja(a)askmonty.org-20100311150203-mg6478pobnln5x22 parent: psergey(a)askmonty.org-20091202142609-18bp41q8mejxl47t committer: sanja(a)askmonty.org branch nick: work-maria-5.2-engine timestamp: Thu 2010-03-11 17:02:03 +0200 message: Maria WL#61 Interface for maria extensions. Alternative plugin interface with additional info (maturity and string version). === modified file 'CMakeLists.txt' --- a/CMakeLists.txt 2009-10-03 19:24:13 +0000 +++ b/CMakeLists.txt 2010-03-11 15:02:03 +0000 @@ -250,7 +250,7 @@ ENDIF(WITH_${ENGINE}_STORAGE_ENGINE AND MYSQL_PLUGIN_STATIC) IF (ENGINE_BUILD_TYPE STREQUAL "STATIC") - SET (mysql_plugin_defs "${mysql_plugin_defs},builtin_${PLUGIN_NAME}_plugin") + SET (maria_plugin_defs "${maria_plugin_defs},builtin_maria_${PLUGIN_NAME}_plugin") SET (MYSQLD_STATIC_ENGINE_LIBS ${MYSQLD_STATIC_ENGINE_LIBS} ${PLUGIN_NAME}) SET (STORAGE_ENGINE_DEFS "${STORAGE_ENGINE_DEFS} -DWITH_${ENGINE}_STORAGE_ENGINE") SET (WITH_${ENGINE}_STORAGE_ENGINE TRUE) @@ -268,7 +268,7 @@ # Special handling for partition(not really pluggable) IF(NOT WITHOUT_PARTITION_STORAGE_ENGINE) SET (STORAGE_ENGINE_DEFS "${STORAGE_ENGINE_DEFS} -DWITH_PARTITION_STORAGE_ENGINE") - SET (mysql_plugin_defs "${mysql_plugin_defs},builtin_partition_plugin") + SET (maria_plugin_defs "${maria_plugin_defs},builtin_maria_partition_plugin") ENDIF(NOT WITHOUT_PARTITION_STORAGE_ENGINE) # Special handling for tmp tables with the maria engine === modified file 'config/ac-macros/plugins.m4' --- a/config/ac-macros/plugins.m4 2009-04-25 10:05:32 +0000 +++ b/config/ac-macros/plugins.m4 2010-03-11 15:02:03 +0000 @@ -460,7 +460,7 @@ ]) ]) ]) - mysql_plugin_defs="$mysql_plugin_defs, [builtin_]$2[_plugin]" + maria_plugin_defs="$maria_plugin_defs, [builtin_maria_]$2[_plugin]" [with_plugin_]$2=yes AC_MSG_RESULT([yes]) m4_ifdef([$11],[ === modified file 'configure.in' --- a/configure.in 2009-11-12 04:31:28 +0000 +++ b/configure.in 2010-03-11 15:02:03 +0000 @@ -2841,7 +2841,7 @@ AC_SUBST(mysql_plugin_dirs) AC_SUBST(mysql_plugin_libs) -AC_SUBST(mysql_plugin_defs) +AC_SUBST(maria_plugin_defs) # Now that sql_client_dirs and sql_server_dirs are stable, determine the union. === modified file 'include/mysql/plugin.h' --- a/include/mysql/plugin.h 2009-09-07 20:50:10 +0000 +++ b/include/mysql/plugin.h 2010-03-11 15:02:03 +0000 @@ -65,7 +65,10 @@ Plugin API. Common for all plugin types. */ +/* MySQL plugin interface version */ #define MYSQL_PLUGIN_INTERFACE_VERSION 0x0100 +/* MariaDB plugin interface version */ +#define MARIA_PLUGIN_INTERFACE_VERSION 0x0100 /* The allowable types of plugins @@ -86,6 +89,21 @@ #define PLUGIN_LICENSE_GPL_STRING "GPL" #define PLUGIN_LICENSE_BSD_STRING "BSD" +/* definitions of code maturity for plugins */ +#define PLUGIN_MATURITY_UNKNOWN 0 +#define PLUGIN_MATURITY_TEST 1 +#define PLUGIN_MATURITY_ALPHA 2 +#define PLUGIN_MATURITY_BETA 3 +#define PLUGIN_MATURITY_GAMMA 4 +#define PLUGIN_MATURITY_RELEASE 5 + +#define PLUGIN_MATURITY_UNKNOWN_STR "Unknown" +#define PLUGIN_MATURITY_TEST_STR "Test" +#define PLUGIN_MATURITY_ALPHA_STR "Alpha" +#define PLUGIN_MATURITY_BETA_STR "Beta" +#define PLUGIN_MATURITY_GAMMA_STR "Gamma" +#define PLUGIN_MATURITY_RELEASE_STR "Release" + /* Macros for beginning and ending plugin declarations. Between mysql_declare_plugin and mysql_declare_plugin_end there should @@ -94,15 +112,29 @@ #ifndef MYSQL_DYNAMIC_PLUGIN + #define __MYSQL_DECLARE_PLUGIN(NAME, VERSION, PSIZE, DECLS) \ int VERSION= MYSQL_PLUGIN_INTERFACE_VERSION; \ int PSIZE= sizeof(struct st_mysql_plugin); \ struct st_mysql_plugin DECLS[]= { + +#define __MARIA_DECLARE_PLUGIN(NAME, VERSION, PSIZE, DECLS) \ +int VERSION= MARIA_PLUGIN_INTERFACE_VERSION; \ +int PSIZE= sizeof(struct st_maria_plugin); \ +struct st_maria_plugin DECLS[]= { + #else + #define __MYSQL_DECLARE_PLUGIN(NAME, VERSION, PSIZE, DECLS) \ MYSQL_PLUGIN_EXPORT int _mysql_plugin_interface_version_= MYSQL_PLUGIN_INTERFACE_VERSION; \ MYSQL_PLUGIN_EXPORT int _mysql_sizeof_struct_st_plugin_= sizeof(struct st_mysql_plugin); \ MYSQL_PLUGIN_EXPORT struct st_mysql_plugin _mysql_plugin_declarations_[]= { + +#define __MARIA_DECLARE_PLUGIN(NAME, VERSION, PSIZE, DECLS) \ +MYSQL_PLUGIN_EXPORT int _maria_plugin_interface_version_= MARIA_PLUGIN_INTERFACE_VERSION; \ +MYSQL_PLUGIN_EXPORT int _maria_sizeof_struct_st_plugin_= sizeof(struct st_maria_plugin); \ +MYSQL_PLUGIN_EXPORT struct st_maria_plugin _maria_plugin_declarations_[]= { + #endif #define mysql_declare_plugin(NAME) \ @@ -111,7 +143,14 @@ builtin_ ## NAME ## _sizeof_struct_st_plugin, \ builtin_ ## NAME ## _plugin) +#define maria_declare_plugin(NAME) \ +__MARIA_DECLARE_PLUGIN(NAME, \ + builtin_maria_ ## NAME ## _plugin_interface_version, \ + builtin_maria_ ## NAME ## _sizeof_struct_st_plugin, \ + builtin_maria_ ## NAME ## _plugin) + #define mysql_declare_plugin_end ,{0,0,0,0,0,0,0,0,0,0,0,0}} +#define maria_declare_plugin_end ,{0,0,0,0,0,0,0,0,0,0,0,0,0,0}} /* declarations for SHOW STATUS support in plugins @@ -407,6 +446,31 @@ void * __reserved1; /* reserved for dependency checking */ }; +/* + MariaDB extension for plugins declaration structure. + + It also copy current MySQL plugin fields to have more independency + in plugins extension +*/ + +struct st_maria_plugin +{ + int type; /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + void *info; /* pointer to type-specific plugin descriptor */ + const char *name; /* plugin name */ + const char *author; /* plugin author (for SHOW PLUGINS) */ + const char *descr; /* general descriptive text (for SHOW PLUGINS ) */ + int license; /* the plugin license (PLUGIN_LICENSE_XXX) */ + int (*init)(void *); /* the function to invoke when plugin is loaded */ + int (*deinit)(void *);/* the function to invoke when plugin is unloaded */ + unsigned int version; /* plugin version (for SHOW PLUGINS) */ + struct st_mysql_show_var *status_vars; + struct st_mysql_sys_var **system_vars; + const char *version_info; /* plugin version string */ + int maturity; /* HA_PLUGIN_MATURITY_XXX */ + void * __reserved1; /* reserved for dependency checking */ +}; + /************************************************************************* API for Full-text parser plugin. (MYSQL_FTPARSER_PLUGIN) */ === modified file 'include/mysql/plugin.h.pp' --- a/include/mysql/plugin.h.pp 2008-10-10 15:28:41 +0000 +++ b/include/mysql/plugin.h.pp 2010-03-11 15:02:03 +0000 @@ -46,6 +46,23 @@ struct st_mysql_sys_var **system_vars; void * __reserved1; }; +struct st_maria_plugin +{ + int type; + void *info; + const char *name; + const char *author; + const char *descr; + int license; + int (*init)(void *); + int (*deinit)(void *); + unsigned int version; + struct st_mysql_show_var *status_vars; + struct st_mysql_sys_var **system_vars; + const char *version_info; + int maturity; + void * __reserved1; +}; enum enum_ftparser_mode { MYSQL_FTPARSER_SIMPLE_MODE= 0, === modified file 'mysql-test/r/information_schema.result' --- a/mysql-test/r/information_schema.result 2009-10-19 17:14:48 +0000 +++ b/mysql-test/r/information_schema.result 2010-03-11 15:02:03 +0000 @@ -1175,7 +1175,7 @@ group by column_type order by num; column_type group_concat(table_schema, '.', table_name) num varchar(27) information_schema.COLUMNS 1 -varchar(7) information_schema.ROUTINES,information_schema.VIEWS 2 +varchar(7) information_schema.PLUGINS,information_schema.ROUTINES,information_schema.VIEWS 3 varchar(20) information_schema.FILES,information_schema.FILES,information_schema.PLUGINS,information_schema.PLUGINS,information_schema.PLUGINS,information_schema.PROFILING 6 create table t1(f1 char(1) not null, f2 char(9) not null) default character set utf8; === modified file 'plugin/daemon_example/daemon_example.cc' --- a/plugin/daemon_example/daemon_example.cc 2007-06-27 14:49:12 +0000 +++ b/plugin/daemon_example/daemon_example.cc 2010-03-11 15:02:03 +0000 @@ -200,3 +200,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(daemon_example) +{ + MYSQL_DAEMON_PLUGIN, + &daemon_example_plugin, + "daemon_example", + "Brian Aker", + "Daemon example, creates a heartbeat beat file in mysql-heartbeat.log", + PLUGIN_LICENSE_GPL, + daemon_example_plugin_init, /* Plugin Init */ + daemon_example_plugin_deinit, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_TEST, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'plugin/fulltext/plugin_example.c' --- a/plugin/fulltext/plugin_example.c 2007-04-26 19:26:04 +0000 +++ b/plugin/fulltext/plugin_example.c 2010-03-11 15:02:03 +0000 @@ -270,4 +270,22 @@ NULL } mysql_declare_plugin_end; +maria_declare_plugin(ftexample) +{ + MYSQL_FTPARSER_PLUGIN, /* type */ + &simple_parser_descriptor, /* descriptor */ + "simple_parser", /* name */ + "MySQL AB", /* author */ + "Simple Full-Text Parser", /* description */ + PLUGIN_LICENSE_GPL, + simple_parser_plugin_init, /* init function (when loaded) */ + simple_parser_plugin_deinit,/* deinit function (when unloaded) */ + 0x0001, /* version */ + simple_status, /* status variables */ + simple_system_variables, /* system variables */ + "0.01", /* string version */ + PLUGIN_MATURITY_TEST, /* maturity */ + NULL +} +maria_declare_plugin_end; === modified file 'sql/ha_ndbcluster.cc' --- a/sql/ha_ndbcluster.cc 2009-09-07 20:50:10 +0000 +++ b/sql/ha_ndbcluster.cc 2010-03-11 15:02:03 +0000 @@ -10561,5 +10561,23 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(ndbcluster) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &ndbcluster_storage_engine, + ndbcluster_hton_name, + "MySQL AB", + "Clustered, fault-tolerant tables", + PLUGIN_LICENSE_GPL, + ndbcluster_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + ndb_status_variables_export,/* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_BETA, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; #endif === modified file 'sql/ha_partition.cc' --- a/sql/ha_partition.cc 2009-11-12 04:31:28 +0000 +++ b/sql/ha_partition.cc 2010-03-11 15:02:03 +0000 @@ -6510,5 +6510,23 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(partition) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &partition_storage_engine, + "partition", + "Mikael Ronstrom, MySQL AB", + "Partition Storage Engine Helper", + PLUGIN_LICENSE_GPL, + partition_initialize, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100, /* 1.0 */ + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; #endif === modified file 'sql/log.cc' --- a/sql/log.cc 2009-11-12 04:31:28 +0000 +++ b/sql/log.cc 2010-03-11 15:02:03 +0000 @@ -5795,3 +5795,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(binlog) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &binlog_storage_engine, + "binlog", + "MySQL AB", + "This is a pseudo storage engine to represent the binlog in a transaction", + PLUGIN_LICENSE_GPL, + binlog_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'sql/sql_builtin.cc.in' --- a/sql/sql_builtin.cc.in 2006-12-31 01:29:11 +0000 +++ b/sql/sql_builtin.cc.in 2010-03-11 15:02:03 +0000 @@ -15,13 +15,12 @@ #include <mysql/plugin.h> -typedef struct st_mysql_plugin builtin_plugin[]; - -extern builtin_plugin - builtin_binlog_plugin@mysql_plugin_defs@; - -struct st_mysql_plugin *mysqld_builtins[]= +typedef struct st_maria_plugin builtin_maria_plugin[]; + +extern builtin_maria_plugin + builtin_maria_binlog_plugin@maria_plugin_defs@; + +struct st_maria_plugin *mariadb_builtins[]= { - builtin_binlog_plugin@mysql_plugin_defs@,(struct st_mysql_plugin *)0 + builtin_maria_binlog_plugin@maria_plugin_defs@,(struct st_maria_plugin *)0 }; - === modified file 'sql/sql_plugin.cc' --- a/sql/sql_plugin.cc 2009-11-12 04:31:28 +0000 +++ b/sql/sql_plugin.cc 2010-03-11 15:02:03 +0000 @@ -27,7 +27,7 @@ #define plugin_int_to_ref(A) &(A) #endif -extern struct st_mysql_plugin *mysqld_builtins[]; +extern struct st_maria_plugin *mariadb_builtins[]; /** @note The order of the enumeration is critical. @@ -82,6 +82,14 @@ "_mysql_sizeof_struct_st_plugin_"; static const char *plugin_declarations_sym= "_mysql_plugin_declarations_"; static int min_plugin_interface_version= MYSQL_PLUGIN_INTERFACE_VERSION & ~0xFF; +static const char *maria_plugin_interface_version_sym= + "_maria_plugin_interface_version_"; +static const char *maria_sizeof_st_plugin_sym= + "_maria_sizeof_struct_st_plugin_"; +static const char *maria_plugin_declarations_sym= + "_maria_plugin_declarations_"; +static int min_maria_plugin_interface_version= + MARIA_PLUGIN_INTERFACE_VERSION & ~0xFF; #endif /* Note that 'int version' must be the first field of every plugin @@ -205,7 +213,7 @@ const char *list); static int test_plugin_options(MEM_ROOT *, struct st_plugin_int *, int *, char **); -static bool register_builtin(struct st_mysql_plugin *, struct st_plugin_int *, +static bool register_builtin(struct st_maria_plugin *, struct st_plugin_int *, struct st_plugin_int **); static void unlock_variables(THD *thd, struct system_variables *vars); static void cleanup_variables(THD *thd, struct system_variables *vars); @@ -341,11 +349,261 @@ dlclose(p->handle); #endif my_free(p->dl.str, MYF(MY_ALLOW_ZERO_PTR)); - if (p->version != MYSQL_PLUGIN_INTERFACE_VERSION) + if (p->mariaversion != MARIA_PLUGIN_INTERFACE_VERSION) my_free((uchar*)p->plugins, MYF(MY_ALLOW_ZERO_PTR)); } +/** + Reads data from mysql plugin interface + + @param plugin_dl Structure where the data should be put + @param sym Reverence on version info + @param dlpath Path to the module + @param report What errors should be reported + + @retval FALSE OK + @retval TRUE ERROR +*/ + +static my_bool read_mysql_plugin_info(struct st_plugin_dl *plugin_dl, + void *sym, char *dlpath, + int report) +{ + DBUG_ENTER("read_maria_plugin_info"); + /* Determine interface version */ + if (!sym) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_interface_version_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_interface_version_sym); + DBUG_RETURN(TRUE); + } + plugin_dl->mariaversion= 0; + plugin_dl->mysqlversion= *(int *)sym; + /* Versioning */ + if (plugin_dl->mysqlversion < min_plugin_interface_version || + (plugin_dl->mysqlversion >> 8) > (MYSQL_PLUGIN_INTERFACE_VERSION >> 8)) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_OPEN_LIBRARY, MYF(0), dlpath, 0, + "plugin interface version mismatch"); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_OPEN_LIBRARY), dlpath, 0, + "plugin interface version mismatch"); + DBUG_RETURN(TRUE); + } + /* Find plugin declarations */ + if (!(sym= dlsym(plugin_dl->handle, plugin_declarations_sym))) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_declarations_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_declarations_sym); + DBUG_RETURN(TRUE); + } + + /* convert mysql declaration to maria one */ + { + int i; + uint sizeof_st_plugin; + struct st_mysql_plugin *old; + struct st_maria_plugin *cur; + char *ptr= (char *)sym; + + if ((sym= dlsym(plugin_dl->handle, sizeof_st_plugin_sym))) + sizeof_st_plugin= *(int *)sym; + else + { +#ifdef ERROR_ON_NO_SIZEOF_PLUGIN_SYMBOL + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), sizeof_st_plugin_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), sizeof_st_plugin_sym); + DBUG_RETURN(TRUE); +#else + /* + When the following assert starts failing, we'll have to switch + to the upper branch of the #ifdef + */ + DBUG_ASSERT(min_plugin_interface_version == 0); + sizeof_st_plugin= (int)offsetof(struct st_mysql_plugin, version); +#endif + } + + for (i= 0; + ((struct st_mysql_plugin *)(ptr+i*sizeof_st_plugin))->info; + i++) + /* no op */; + + cur= (struct st_maria_plugin*) + my_malloc(i * sizeof(struct st_maria_plugin), + MYF(MY_ZEROFILL|MY_WME)); + if (!cur) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_OUTOFMEMORY, MYF(0), plugin_dl->dl.length); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_OUTOFMEMORY), plugin_dl->dl.length); + DBUG_RETURN(TRUE); + } + /* + All st_plugin fields not initialized in the plugin explicitly, are + set to 0. It matches C standard behaviour for struct initializers that + have less values than the struct definition. + */ + for (i=0; + (old=(struct st_mysql_plugin *)(ptr+i*sizeof_st_plugin))->info; + i++) + { + + cur->type= old->type; + cur->info= old->info; + cur->name= old->name; + cur->author= old->author; + cur->descr= old->descr; + cur->license= old->license; + cur->init= old->init; + cur->deinit= old->deinit; + cur->version= old->version; + cur->status_vars= old->status_vars; + cur->system_vars= old->system_vars; + /* + Something like this should be added to process + new mysql plugin versions: + if (plugin_dl->mysqlversion > 0x0100) + { + cur->newfield= CONSTANT_MEANS_UNKNOWN; + } + else + { + cur->newfield= old->newfield; + } + */ + /* Maria only fields */ + cur->version_info= "Unknown"; + cur->maturity= PLUGIN_MATURITY_UNKNOWN; + } + + plugin_dl->plugins= (struct st_maria_plugin *)cur; + } + + DBUG_RETURN(FALSE); +} + + +/** + Reads data from maria plugin interface + + @param plugin_dl Structure where the data should be put + @param sym Reverence on version info + @param dlpath Path to the module + @param report what errors should be reported + + @retval FALSE OK + @retval TRUE ERROR +*/ + +static my_bool read_maria_plugin_info(struct st_plugin_dl *plugin_dl, + void *sym, char *dlpath, + int report) +{ + DBUG_ENTER("read_maria_plugin_info"); + + /* Determine interface version */ + if (!(sym)) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_interface_version_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_interface_version_sym); + DBUG_RETURN(TRUE); + } + plugin_dl->mariaversion= *(int *)sym; + plugin_dl->mysqlversion= 0; + /* Versioning */ + if (plugin_dl->mariaversion < min_maria_plugin_interface_version || + (plugin_dl->mariaversion >> 8) > (MARIA_PLUGIN_INTERFACE_VERSION >> 8)) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_OPEN_LIBRARY, MYF(0), dlpath, 0, + "plugin interface version mismatch"); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_OPEN_LIBRARY), dlpath, 0, + "plugin interface version mismatch"); + DBUG_RETURN(TRUE); + } + /* Find plugin declarations */ + if (!(sym= dlsym(plugin_dl->handle, maria_plugin_declarations_sym))) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_declarations_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_declarations_sym); + DBUG_RETURN(TRUE); + } + if (plugin_dl->mariaversion != MARIA_PLUGIN_INTERFACE_VERSION) + { + int i; + uint sizeof_st_plugin; + struct st_maria_plugin *old, *cur; + char *ptr= (char *)sym; + + if ((sym= dlsym(plugin_dl->handle, maria_sizeof_st_plugin_sym))) + sizeof_st_plugin= *(int *)sym; + else + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), sizeof_st_plugin_sym); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), sizeof_st_plugin_sym); + DBUG_RETURN(TRUE); + } + + for (i= 0; + ((struct st_maria_plugin *)(ptr+i*sizeof_st_plugin))->info; + i++) + /* no op */; + + cur= (struct st_maria_plugin*) + my_malloc(i * sizeof(struct st_maria_plugin), + MYF(MY_ZEROFILL|MY_WME)); + if (!cur) + { + free_plugin_mem(plugin_dl); + if (report & REPORT_TO_USER) + my_error(ER_OUTOFMEMORY, MYF(0), plugin_dl->dl.length); + if (report & REPORT_TO_LOG) + sql_print_error(ER(ER_OUTOFMEMORY), plugin_dl->dl.length); + DBUG_RETURN(TRUE); + } + /* + All st_plugin fields not initialized in the plugin explicitly, are + set to 0. It matches C standard behaviour for struct initializers that + have less values than the struct definition. + */ + for (i=0; + (old=(struct st_maria_plugin *)(ptr+i*sizeof_st_plugin))->info; + i++) + memcpy(cur+i, old, min(sizeof(cur[i]), sizeof_st_plugin)); + + sym= cur; + } + plugin_dl->plugins= (struct st_maria_plugin *)sym; + + DBUG_RETURN(FALSE); +} + static st_plugin_dl *plugin_dl_add(const LEX_STRING *dl, int report) { #ifdef HAVE_DLOPEN @@ -399,98 +657,22 @@ sql_print_error(ER(ER_CANT_OPEN_LIBRARY), dlpath, errno, errmsg); DBUG_RETURN(0); } - /* Determine interface version */ - if (!(sym= dlsym(plugin_dl.handle, plugin_interface_version_sym))) - { - free_plugin_mem(&plugin_dl); - if (report & REPORT_TO_USER) - my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_interface_version_sym); - if (report & REPORT_TO_LOG) - sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_interface_version_sym); - DBUG_RETURN(0); - } - plugin_dl.version= *(int *)sym; - /* Versioning */ - if (plugin_dl.version < min_plugin_interface_version || - (plugin_dl.version >> 8) > (MYSQL_PLUGIN_INTERFACE_VERSION >> 8)) - { - free_plugin_mem(&plugin_dl); - if (report & REPORT_TO_USER) - my_error(ER_CANT_OPEN_LIBRARY, MYF(0), dlpath, 0, - "plugin interface version mismatch"); - if (report & REPORT_TO_LOG) - sql_print_error(ER(ER_CANT_OPEN_LIBRARY), dlpath, 0, - "plugin interface version mismatch"); - DBUG_RETURN(0); - } - /* Find plugin declarations */ - if (!(sym= dlsym(plugin_dl.handle, plugin_declarations_sym))) - { - free_plugin_mem(&plugin_dl); - if (report & REPORT_TO_USER) - my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), plugin_declarations_sym); - if (report & REPORT_TO_LOG) - sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), plugin_declarations_sym); - DBUG_RETURN(0); - } - - if (plugin_dl.version != MYSQL_PLUGIN_INTERFACE_VERSION) - { - int i; - uint sizeof_st_plugin; - struct st_mysql_plugin *old, *cur; - char *ptr= (char *)sym; - - if ((sym= dlsym(plugin_dl.handle, sizeof_st_plugin_sym))) - sizeof_st_plugin= *(int *)sym; - else - { -#ifdef ERROR_ON_NO_SIZEOF_PLUGIN_SYMBOL - free_plugin_mem(&plugin_dl); - if (report & REPORT_TO_USER) - my_error(ER_CANT_FIND_DL_ENTRY, MYF(0), sizeof_st_plugin_sym); - if (report & REPORT_TO_LOG) - sql_print_error(ER(ER_CANT_FIND_DL_ENTRY), sizeof_st_plugin_sym); - DBUG_RETURN(0); -#else - /* - When the following assert starts failing, we'll have to switch - to the upper branch of the #ifdef - */ - DBUG_ASSERT(min_plugin_interface_version == 0); - sizeof_st_plugin= (int)offsetof(struct st_mysql_plugin, version); -#endif - } - - for (i= 0; - ((struct st_mysql_plugin *)(ptr+i*sizeof_st_plugin))->info; - i++) - /* no op */; - - cur= (struct st_mysql_plugin*) - my_malloc(i*sizeof(struct st_mysql_plugin), MYF(MY_ZEROFILL|MY_WME)); - if (!cur) - { - free_plugin_mem(&plugin_dl); - if (report & REPORT_TO_USER) - my_error(ER_OUTOFMEMORY, MYF(0), plugin_dl.dl.length); - if (report & REPORT_TO_LOG) - sql_print_error(ER(ER_OUTOFMEMORY), plugin_dl.dl.length); - DBUG_RETURN(0); - } - /* - All st_plugin fields not initialized in the plugin explicitly, are - set to 0. It matches C standard behaviour for struct initializers that - have less values than the struct definition. - */ - for (i=0; - (old=(struct st_mysql_plugin *)(ptr+i*sizeof_st_plugin))->info; - i++) - memcpy(cur+i, old, min(sizeof(cur[i]), sizeof_st_plugin)); - - sym= cur; - } - plugin_dl.plugins= (struct st_mysql_plugin *)sym; + + /* Checks which plugin interface present and reads info */ + if (!(sym= dlsym(plugin_dl.handle, maria_plugin_interface_version_sym))) + { + if (read_mysql_plugin_info(&plugin_dl, + dlsym(plugin_dl.handle, + plugin_interface_version_sym), + dlpath, + report)) + DBUG_RETURN(0); + } + else + { + if (read_maria_plugin_info(&plugin_dl, sym, dlpath, report)) + DBUG_RETURN(0); + } /* Duplicate and convert dll name */ plugin_dl.dl.length= dl->length * files_charset_info->mbmaxlen + 1; @@ -718,7 +900,7 @@ int *argc, char **argv, int report) { struct st_plugin_int tmp; - struct st_mysql_plugin *plugin; + struct st_maria_plugin *plugin; DBUG_ENTER("plugin_add"); if (plugin_find_internal(name, MYSQL_ANY_PLUGIN)) { @@ -1120,8 +1302,8 @@ { uint i; bool is_myisam; - struct st_mysql_plugin **builtins; - struct st_mysql_plugin *plugin; + struct st_maria_plugin **builtins; + struct st_maria_plugin *plugin; struct st_plugin_int tmp, *plugin_ptr, **reap; MEM_ROOT tmp_root; bool reaped_mandatory_plugin= FALSE; @@ -1160,7 +1342,7 @@ /* First we register builtin plugins */ - for (builtins= mysqld_builtins; *builtins; builtins++) + for (builtins= mariadb_builtins; *builtins; builtins++) { for (plugin= *builtins; plugin->info; plugin++) { @@ -1290,7 +1472,7 @@ } -static bool register_builtin(struct st_mysql_plugin *plugin, +static bool register_builtin(struct st_maria_plugin *plugin, struct st_plugin_int *tmp, struct st_plugin_int **ptr) { @@ -1326,7 +1508,7 @@ RETURN false - plugin registered successfully */ -bool plugin_register_builtin(THD *thd, struct st_mysql_plugin *plugin) +bool plugin_register_builtin(THD *thd, struct st_maria_plugin *plugin) { struct st_plugin_int tmp, *ptr; bool result= true; @@ -1455,7 +1637,7 @@ char buffer[FN_REFLEN]; LEX_STRING name= {buffer, 0}, dl= {NULL, 0}, *str= &name; struct st_plugin_dl *plugin_dl; - struct st_mysql_plugin *plugin; + struct st_maria_plugin *plugin; char *p= buffer; DBUG_ENTER("plugin_load_list"); while (list) === modified file 'sql/sql_plugin.h' --- a/sql/sql_plugin.h 2009-05-14 12:03:33 +0000 +++ b/sql/sql_plugin.h 2010-03-11 15:02:03 +0000 @@ -62,8 +62,9 @@ { LEX_STRING dl; void *handle; - struct st_mysql_plugin *plugins; - int version; + struct st_maria_plugin *plugins; + int mysqlversion; + int mariaversion; uint ref_count; /* number of plugins loaded from the library */ }; @@ -72,7 +73,7 @@ struct st_plugin_int { LEX_STRING name; - struct st_mysql_plugin *plugin; + struct st_maria_plugin *plugin; struct st_plugin_dl *plugin_dl; uint state; uint ref_count; /* number of threads using the plugin */ === modified file 'sql/sql_show.cc' --- a/sql/sql_show.cc 2009-11-12 04:31:28 +0000 +++ b/sql/sql_show.cc 2010-03-11 15:02:03 +0000 @@ -94,11 +94,19 @@ return my_snprintf(buf, buf_length, "%d.%d", version>>8,version&0xff); } +static const LEX_STRING maturity_name[]={ + { C_STRING_WITH_LEN(PLUGIN_MATURITY_UNKNOWN_STR) }, + { C_STRING_WITH_LEN(PLUGIN_MATURITY_TEST_STR) }, + { C_STRING_WITH_LEN(PLUGIN_MATURITY_ALPHA_STR) }, + { C_STRING_WITH_LEN(PLUGIN_MATURITY_BETA_STR) }, + { C_STRING_WITH_LEN(PLUGIN_MATURITY_GAMMA_STR) }, + { C_STRING_WITH_LEN(PLUGIN_MATURITY_RELEASE_STR) }}; + static my_bool show_plugins(THD *thd, plugin_ref plugin, void *arg) { TABLE *table= (TABLE*) arg; - struct st_mysql_plugin *plug= plugin_decl(plugin); + struct st_maria_plugin *plug= plugin_decl(plugin); struct st_plugin_dl *plugin_dl= plugin_dlib(plugin); CHARSET_INFO *cs= system_charset_info; char version_buf[20]; @@ -143,7 +151,9 @@ table->field[5]->set_notnull(); table->field[6]->store(version_buf, make_version_string(version_buf, sizeof(version_buf), - plugin_dl->version), + (plugin_dl->mariaversion ? + plugin_dl->mariaversion : + plugin_dl->mysqlversion)), cs); table->field[6]->set_notnull(); } @@ -186,6 +196,26 @@ } table->field[9]->set_notnull(); + if ((uint) plug->maturity <= PLUGIN_MATURITY_RELEASE) + table->field[10]->store(maturity_name[plug->maturity].str, + maturity_name[plug->maturity].length, + cs); + else + { + DBUG_ASSERT(0); + table->field[10]->store("Unknown", 7, cs); + } + table->field[10]->set_notnull(); + + if (plug->version_info) + { + table->field[11]->store(plug->version_info, + strlen(plug->version_info), cs); + table->field[11]->set_notnull(); + } + else + table->field[11]->set_null(); + return schema_table_store_record(thd, table); } @@ -4293,7 +4323,7 @@ if (plugin_state(plugin) != PLUGIN_IS_READY) { - struct st_mysql_plugin *plug= plugin_decl(plugin); + struct st_maria_plugin *plug= plugin_decl(plugin); if (!(wild && wild[0] && wild_case_compare(scs, plug->name,wild))) { @@ -6990,6 +7020,8 @@ {"PLUGIN_AUTHOR", NAME_CHAR_LEN, MYSQL_TYPE_STRING, 0, 1, 0, SKIP_OPEN_TABLE}, {"PLUGIN_DESCRIPTION", 65535, MYSQL_TYPE_STRING, 0, 1, 0, SKIP_OPEN_TABLE}, {"PLUGIN_LICENSE", 80, MYSQL_TYPE_STRING, 0, 1, "License", SKIP_OPEN_TABLE}, + {"PLUGIN_MATURITY", 7, MYSQL_TYPE_STRING, 0, 1, 0, SKIP_OPEN_TABLE}, + {"PLUGIN_AUTH_VERSION", 80, MYSQL_TYPE_STRING, 0, 1, 0, SKIP_OPEN_TABLE}, {0, 0, MYSQL_TYPE_STRING, 0, 0, 0, SKIP_OPEN_TABLE} }; === modified file 'storage/archive/ha_archive.cc' --- a/storage/archive/ha_archive.cc 2009-09-07 20:50:10 +0000 +++ b/storage/archive/ha_archive.cc 2010-03-11 15:02:03 +0000 @@ -1642,4 +1642,22 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(archive) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &archive_storage_engine, + "ARCHIVE", + "Brian Aker, MySQL AB", + "Archive storage engine", + PLUGIN_LICENSE_GPL, + archive_db_init, /* Plugin Init */ + archive_db_done, /* Plugin Deinit */ + 0x0300 /* 3.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/blackhole/ha_blackhole.cc' --- a/storage/blackhole/ha_blackhole.cc 2008-11-10 20:21:49 +0000 +++ b/storage/blackhole/ha_blackhole.cc 2010-03-11 15:02:03 +0000 @@ -369,3 +369,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(blackhole) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &blackhole_storage_engine, + "BLACKHOLE", + "MySQL AB", + "/dev/null storage engine (anything you write to it disappears)", + PLUGIN_LICENSE_GPL, + blackhole_init, /* Plugin Init */ + blackhole_fini, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/csv/ha_tina.cc' --- a/storage/csv/ha_tina.cc 2009-04-25 10:05:32 +0000 +++ b/storage/csv/ha_tina.cc 2010-03-11 15:02:03 +0000 @@ -1636,4 +1636,21 @@ NULL /* config options */ } mysql_declare_plugin_end; - +maria_declare_plugin(csv) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &csv_storage_engine, + "CSV", + "Brian Aker, MySQL AB", + "CSV storage engine", + PLUGIN_LICENSE_GPL, + tina_init_func, /* Plugin Init */ + tina_done_func, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/example/ha_example.cc' --- a/storage/example/ha_example.cc 2008-02-24 13:12:17 +0000 +++ b/storage/example/ha_example.cc 2010-03-11 15:02:03 +0000 @@ -906,3 +906,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(example) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &example_storage_engine, + "EXAMPLE", + "Brian Aker, MySQL AB", + "Example storage engine", + PLUGIN_LICENSE_GPL, + example_init_func, /* Plugin Init */ + example_done_func, /* Plugin Deinit */ + 0x0001 /* 0.1 */, + NULL, /* status variables */ + example_system_variables, /* system variables */ + "0.1", /* string version */ + PLUGIN_MATURITY_TEST, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/federated/ha_federated.cc' --- a/storage/federated/ha_federated.cc 2009-09-07 20:50:10 +0000 +++ b/storage/federated/ha_federated.cc 2010-03-11 15:02:03 +0000 @@ -3379,3 +3379,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(federated) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &federated_storage_engine, + "FEDERATED", + "Patrick Galbraith and Brian Aker, MySQL AB", + "Federated MySQL storage engine", + PLUGIN_LICENSE_GPL, + federated_db_init, /* Plugin Init */ + federated_done, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_BETA, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/federatedx/ha_federatedx.cc' --- a/storage/federatedx/ha_federatedx.cc 2009-11-03 11:08:09 +0000 +++ b/storage/federatedx/ha_federatedx.cc 2010-03-11 15:02:03 +0000 @@ -3485,9 +3485,27 @@ PLUGIN_LICENSE_GPL, federatedx_db_init, /* Plugin Init */ federatedx_done, /* Plugin Deinit */ - 0x0100 /* 1.0 */, + 0x0200 /* 2.0 */, NULL, /* status variables */ NULL, /* system variables */ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(federated) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &federatedx_storage_engine, + "FEDERATED", + "Patrick Galbraith", + "FederatedX pluggable storage engine", + PLUGIN_LICENSE_GPL, + federatedx_db_init, /* Plugin Init */ + federatedx_done, /* Plugin Deinit */ + 0x0200 /* 2.0 */, + NULL, /* status variables */ + NULL, /* system variables */ + "2.0", /* string version */ + PLUGIN_MATURITY_BETA, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/heap/ha_heap.cc' --- a/storage/heap/ha_heap.cc 2009-09-07 20:50:10 +0000 +++ b/storage/heap/ha_heap.cc 2010-03-11 15:02:03 +0000 @@ -767,3 +767,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(heap) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &heap_storage_engine, + "MEMORY", + "MySQL AB", + "Hash based, stored in memory, useful for temporary tables", + PLUGIN_LICENSE_GPL, + heap_init, + NULL, + 0x0100, /* 1.0 */ + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/ibmdb2i/ha_ibmdb2i.cc' --- a/storage/ibmdb2i/ha_ibmdb2i.cc 2009-07-08 09:10:01 +0000 +++ b/storage/ibmdb2i/ha_ibmdb2i.cc 2010-03-11 15:02:03 +0000 @@ -3357,3 +3357,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(ibmdb2i) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &ibmdb2i_storage_engine, + "IBMDB2I", + "The IBM development team in Rochester, Minnesota", + "IBM DB2 for i Storage Engine", + PLUGIN_LICENSE_GPL, + ibmdb2i_init_func, /* Plugin Init */ + ibmdb2i_done_func, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + NULL, /* status variables */ + ibmdb2i_system_variables, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_UNKNOWN, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/innobase/handler/ha_innodb.cc' --- a/storage/innobase/handler/ha_innodb.cc 2009-10-16 22:57:48 +0000 +++ b/storage/innobase/handler/ha_innodb.cc 2010-03-11 15:02:03 +0000 @@ -8684,6 +8684,24 @@ NULL /* reserved */ } mysql_declare_plugin_end; +maria_declare_plugin(innobase) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &innobase_storage_engine, + innobase_hton_name, + "Innobase OY", + "Supports transactions, row-level locking, and foreign keys", + PLUGIN_LICENSE_GPL, + innobase_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100 /* 1.0 */, + innodb_status_variables_export,/* status variables */ + innobase_system_variables, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* reserved */ +} +maria_declare_plugin_end; /** @brief Initialize the default value of innodb_commit_concurrency. === modified file 'storage/innodb_plugin/handler/i_s.cc' --- a/storage/innodb_plugin/handler/i_s.cc 2009-08-14 15:18:52 +0000 +++ b/storage/innodb_plugin/handler/i_s.cc 2010-03-11 15:02:03 +0000 @@ -455,6 +455,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_trx_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_TRX"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB transactions"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_trx_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table INFORMATION_SCHEMA.innodb_locks */ static ST_FIELD_INFO innodb_locks_fields_info[] = { @@ -730,6 +787,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_locks_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_LOCKS"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB conflicting locks"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_locks_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table INFORMATION_SCHEMA.innodb_lock_waits */ static ST_FIELD_INFO innodb_lock_waits_fields_info[] = { @@ -913,6 +1027,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_lock_waits_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_LOCK_WAITS"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, "Innobase Oy"), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB which lock is blocking which"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_lock_waits_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /*******************************************************************//** Common function to fill any of the dynamic tables: INFORMATION_SCHEMA.innodb_trx @@ -1245,6 +1416,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_mysql_plugin i_s_innodb_cmp_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMP"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compression"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmp_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_cmp_reset = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -1295,6 +1523,64 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmp_reset_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMP_RESET"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compression;" + " reset cumulated counts"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmp_reset_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table information_schema.innodb_cmpmem. */ static ST_FIELD_INFO i_s_cmpmem_fields_info[] = { @@ -1511,6 +1797,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmpmem_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMPMEM"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compressed buffer pool"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmpmem_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_cmpmem_reset = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -1561,6 +1904,64 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmpmem_reset_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMPMEM_RESET"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compressed buffer pool;" + " reset cumulated counts"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmpmem_reset_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /*******************************************************************//** Unbind a dynamic INFORMATION_SCHEMA table. @return 0 on success */ === modified file 'storage/maria/ha_maria.cc' --- a/storage/maria/ha_maria.cc 2009-10-26 11:35:42 +0000 +++ b/storage/maria/ha_maria.cc 2010-03-11 15:02:03 +0000 @@ -3346,9 +3346,27 @@ PLUGIN_LICENSE_GPL, ha_maria_init, /* Plugin Init */ NULL, /* Plugin Deinit */ - 0x0100, /* 1.0 */ + 0x0105, /* 1.5 */ status_variables, /* status variables */ system_variables, /* system variables */ NULL } mysql_declare_plugin_end; +maria_declare_plugin(maria) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &maria_storage_engine, + "MARIA", + "MySQL AB", + "Crash-safe tables with MyISAM heritage", + PLUGIN_LICENSE_GPL, + ha_maria_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0105, /* 1.5 */ + status_variables, /* status variables */ + system_variables, /* system variables */ + "1.5", /* string version */ + PLUGIN_MATURITY_GAMMA, /* maturity */ + NULL +} +maria_declare_plugin_end; === modified file 'storage/myisam/ha_myisam.cc' --- a/storage/myisam/ha_myisam.cc 2009-10-17 19:12:28 +0000 +++ b/storage/myisam/ha_myisam.cc 2010-03-11 15:02:03 +0000 @@ -2183,6 +2183,24 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(myisam) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &myisam_storage_engine, + "MyISAM", + "MySQL AB", + "Default engine as of MySQL 3.23 with great performance", + PLUGIN_LICENSE_GPL, + myisam_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100, /* 1.0 */ + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; #ifdef HAVE_QUERY_CACHE === modified file 'storage/myisammrg/ha_myisammrg.cc' --- a/storage/myisammrg/ha_myisammrg.cc 2009-10-15 21:38:29 +0000 +++ b/storage/myisammrg/ha_myisammrg.cc 2010-03-11 15:02:03 +0000 @@ -1289,3 +1289,21 @@ NULL /* config options */ } mysql_declare_plugin_end; +maria_declare_plugin(myisammrg) +{ + MYSQL_STORAGE_ENGINE_PLUGIN, + &myisammrg_storage_engine, + "MRG_MYISAM", + "MySQL AB", + "Collection of identical MyISAM tables", + PLUGIN_LICENSE_GPL, + myisammrg_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + 0x0100, /* 1.0 */ + NULL, /* status variables */ + NULL, /* system variables */ + "1.0", /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; === modified file 'storage/pbxt/src/ha_pbxt.cc' --- a/storage/pbxt/src/ha_pbxt.cc 2009-09-03 06:15:03 +0000 +++ b/storage/pbxt/src/ha_pbxt.cc 2010-03-11 15:02:03 +0000 @@ -5507,6 +5507,42 @@ drizzle_declare_plugin_end; #else mysql_declare_plugin_end; +#ifdef MARIADB_BASE_VERSION +maria_declare_plugin(pbxt) +{ /* PBXT */ + MYSQL_STORAGE_ENGINE_PLUGIN, + &pbxt_storage_engine, + "PBXT", + "Paul McCullagh, PrimeBase Technologies GmbH", + "High performance, multi-versioning transactional engine", + PLUGIN_LICENSE_GPL, + pbxt_init, /* Plugin Init */ + pbxt_end, /* Plugin Deinit */ + 0x0001 /* 0.1 */, + NULL, /* status variables */ + pbxt_system_variables, /* system variables */ + "1.0.09g RC3", /* string version */ + PLUGIN_MATURITY_GAMMA, /* maturity */ + NULL /* config options */ +}, +{ /* PBXT_STATISTICS */ + MYSQL_INFORMATION_SCHEMA_PLUGIN, + &pbxt_statitics, + "PBXT_STATISTICS", + "Paul McCullagh, PrimeBase Technologies GmbH", + "PBXT internal system statitics", + PLUGIN_LICENSE_GPL, + pbxt_init_statitics, /* plugin init */ + pbxt_exit_statitics, /* plugin deinit */ + 0x0005, + NULL, /* status variables */ + NULL, /* system variables */ + "1.0.09g RC3", /* string version */ + PLUGIN_MATURITY_GAMMA, /* maturity */ + NULL /* config options */ +} +maria_declare_plugin_end; +#endif #endif #if defined(XT_WIN) && defined(XT_COREDUMP) === modified file 'storage/xtradb/handler/ha_innodb.cc' --- a/storage/xtradb/handler/ha_innodb.cc 2009-10-16 22:57:48 +0000 +++ b/storage/xtradb/handler/ha_innodb.cc 2010-03-11 15:02:03 +0000 @@ -10540,6 +10540,39 @@ i_s_innodb_index_stats, i_s_innodb_patches mysql_declare_plugin_end; +maria_declare_plugin(innobase) +{ /* InnoDB */ + MYSQL_STORAGE_ENGINE_PLUGIN, + &innobase_storage_engine, + innobase_hton_name, + "Innobase Oy", + "Supports transactions, row-level locking, and foreign keys", + PLUGIN_LICENSE_GPL, + innobase_init, /* Plugin Init */ + NULL, /* Plugin Deinit */ + INNODB_VERSION_SHORT, + innodb_status_variables_export,/* status variables */ + innobase_system_variables, /* system variables */ + INNODB_VERSION_STR, /* string version */ + PLUGIN_MATURITY_RELEASE, /* maturity */ + NULL /* reserved */ +}, +i_s_innodb_rseg_maria, +i_s_innodb_buffer_pool_pages_maria, +i_s_innodb_buffer_pool_pages_index_maria, +i_s_innodb_buffer_pool_pages_blob_maria, +i_s_innodb_trx_maria, +i_s_innodb_locks_maria, +i_s_innodb_lock_waits_maria, +i_s_innodb_cmp_maria, +i_s_innodb_cmp_reset_maria, +i_s_innodb_cmpmem_maria, +i_s_innodb_cmpmem_reset_maria, +i_s_innodb_table_stats_maria, +i_s_innodb_index_stats_maria, +i_s_innodb_patches_maria +maria_declare_plugin_end; + /** @brief Initialize the default value of innodb_commit_concurrency. === modified file 'storage/xtradb/handler/i_s.cc' --- a/storage/xtradb/handler/i_s.cc 2009-09-15 10:46:35 +0000 +++ b/storage/xtradb/handler/i_s.cc 2010-03-11 15:02:03 +0000 @@ -390,6 +390,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_patches_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "XTRADB_ENHANCEMENTS"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, "Percona"), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Enhancements applied to InnoDB plugin"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_patches_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + static ST_FIELD_INFO i_s_innodb_buffer_pool_pages_fields_info[] = { @@ -1037,6 +1094,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_buffer_pool_pages_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_BUFFER_POOL_PAGES"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB buffer pool pages"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_innodb_buffer_pool_pages_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, 0x0100 /* 1.0 */), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_buffer_pool_pages_index = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -1086,6 +1200,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_buffer_pool_pages_index_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_BUFFER_POOL_PAGES_INDEX"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB buffer pool index pages"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_innodb_buffer_pool_pages_index_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, 0x0100 /* 1.0 */), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_buffer_pool_pages_blob = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -1135,6 +1306,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_buffer_pool_pages_blob_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_BUFFER_POOL_PAGES_BLOB"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB buffer pool blob pages"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_innodb_buffer_pool_pages_blob_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, 0x0100 /* 1.0 */), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table INFORMATION_SCHEMA.innodb_trx */ static ST_FIELD_INFO innodb_trx_fields_info[] = @@ -1370,6 +1598,64 @@ STRUCT_FLD(__reserved1, NULL) }; + +UNIV_INTERN struct st_maria_plugin i_s_innodb_trx_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_TRX"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB transactions"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_trx_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table INFORMATION_SCHEMA.innodb_locks */ static ST_FIELD_INFO innodb_locks_fields_info[] = { @@ -1645,6 +1931,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_locks_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_LOCKS"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB conflicting locks"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_locks_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /* Fields of the dynamic table INFORMATION_SCHEMA.innodb_lock_waits */ static ST_FIELD_INFO innodb_lock_waits_fields_info[] = { @@ -1828,6 +2171,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_lock_waits_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_LOCK_WAITS"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, "Innobase Oy"), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB which lock is blocking which"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, innodb_lock_waits_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /*********************************************************************** Common function to fill any of the dynamic tables: INFORMATION_SCHEMA.innodb_trx @@ -2160,6 +2560,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmp_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMP"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compression"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmp_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_cmp_reset = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -2210,6 +2667,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmp_reset_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMP_RESET"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compression;" + " reset cumulated counts"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmp_reset_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; /* Fields of the dynamic table information_schema.innodb_cmpmem. */ static ST_FIELD_INFO i_s_cmpmem_fields_info[] = { @@ -2428,6 +2942,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmpmem_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMPMEM"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compressed buffer pool"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmpmem_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_cmpmem_reset = { /* the plugin type (a MYSQL_XXX_PLUGIN value) */ @@ -2478,6 +3049,64 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_cmpmem_reset_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_CMPMEM_RESET"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "Statistics for the InnoDB compressed buffer pool;" + " reset cumulated counts"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_cmpmem_reset_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, INNODB_VERSION_SHORT), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /*********************************************************************** Unbind a dynamic INFORMATION_SCHEMA table. */ static @@ -2657,6 +3286,63 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_rseg_maria = +{ + /* the plugin type (a MYSQL_XXX_PLUGIN value) */ + /* int */ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + + /* pointer to type-specific plugin descriptor */ + /* void* */ + STRUCT_FLD(info, &i_s_info), + + /* plugin name */ + /* const char* */ + STRUCT_FLD(name, "INNODB_RSEG"), + + /* plugin author (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(author, plugin_author), + + /* general descriptive text (for SHOW PLUGINS) */ + /* const char* */ + STRUCT_FLD(descr, "InnoDB rollback segment information"), + + /* the plugin license (PLUGIN_LICENSE_XXX) */ + /* int */ + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + + /* the function to invoke when plugin is loaded */ + /* int (*)(void*); */ + STRUCT_FLD(init, i_s_innodb_rseg_init), + + /* the function to invoke when plugin is unloaded */ + /* int (*)(void*); */ + STRUCT_FLD(deinit, i_s_common_deinit), + + /* plugin version (for SHOW PLUGINS) */ + /* unsigned int */ + STRUCT_FLD(version, 0x0100 /* 1.0 */), + + /* struct st_mysql_show_var* */ + STRUCT_FLD(status_vars, NULL), + + /* struct st_mysql_sys_var** */ + STRUCT_FLD(system_vars, NULL), + + /* string version */ + /* const char * */ + STRUCT_FLD(version_info, "1.0"), + + /* Maturity */ + /* int */ + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + + /* reserved for dependency checking */ + /* void* */ + STRUCT_FLD(__reserved1, NULL) +}; + /*********************************************************************** */ static ST_FIELD_INFO i_s_innodb_table_stats_info[] = @@ -2937,6 +3623,24 @@ STRUCT_FLD(__reserved1, NULL) }; +UNIV_INTERN struct st_maria_plugin i_s_innodb_table_stats_maria = +{ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + STRUCT_FLD(info, &i_s_info), + STRUCT_FLD(name, "INNODB_TABLE_STATS"), + STRUCT_FLD(author, plugin_author), + STRUCT_FLD(descr, "InnoDB table statistics in memory"), + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + STRUCT_FLD(init, i_s_innodb_table_stats_init), + STRUCT_FLD(deinit, i_s_common_deinit), + STRUCT_FLD(version, 0x0100 /* 1.0 */), + STRUCT_FLD(status_vars, NULL), + STRUCT_FLD(system_vars, NULL), + STRUCT_FLD(version_info, "1.0"), + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + STRUCT_FLD(__reserved1, NULL) +}; + UNIV_INTERN struct st_mysql_plugin i_s_innodb_index_stats = { STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), @@ -2952,3 +3656,21 @@ STRUCT_FLD(system_vars, NULL), STRUCT_FLD(__reserved1, NULL) }; + +UNIV_INTERN struct st_maria_plugin i_s_innodb_index_stats_maria = +{ + STRUCT_FLD(type, MYSQL_INFORMATION_SCHEMA_PLUGIN), + STRUCT_FLD(info, &i_s_info), + STRUCT_FLD(name, "INNODB_INDEX_STATS"), + STRUCT_FLD(author, plugin_author), + STRUCT_FLD(descr, "InnoDB index statistics in memory"), + STRUCT_FLD(license, PLUGIN_LICENSE_GPL), + STRUCT_FLD(init, i_s_innodb_index_stats_init), + STRUCT_FLD(deinit, i_s_common_deinit), + STRUCT_FLD(version, 0x0100 /* 1.0 */), + STRUCT_FLD(status_vars, NULL), + STRUCT_FLD(system_vars, NULL), + STRUCT_FLD(version_info, "1.0"), + STRUCT_FLD(maturity, PLUGIN_MATURITY_RELEASE), + STRUCT_FLD(__reserved1, NULL) +}; === modified file 'storage/xtradb/handler/i_s.h' --- a/storage/xtradb/handler/i_s.h 2009-06-25 01:43:25 +0000 +++ b/storage/xtradb/handler/i_s.h 2010-03-11 15:02:03 +0000 @@ -40,4 +40,19 @@ extern struct st_mysql_plugin i_s_innodb_table_stats; extern struct st_mysql_plugin i_s_innodb_index_stats; +extern struct st_maria_plugin i_s_innodb_buffer_pool_pages_maria; +extern struct st_maria_plugin i_s_innodb_buffer_pool_pages_index_maria; +extern struct st_maria_plugin i_s_innodb_buffer_pool_pages_blob_maria; +extern struct st_maria_plugin i_s_innodb_trx_maria; +extern struct st_maria_plugin i_s_innodb_locks_maria; +extern struct st_maria_plugin i_s_innodb_lock_waits_maria; +extern struct st_maria_plugin i_s_innodb_cmp_maria; +extern struct st_maria_plugin i_s_innodb_cmp_reset_maria; +extern struct st_maria_plugin i_s_innodb_cmpmem_maria; +extern struct st_maria_plugin i_s_innodb_cmpmem_reset_maria; +extern struct st_maria_plugin i_s_innodb_patches_maria; +extern struct st_maria_plugin i_s_innodb_rseg_maria; +extern struct st_maria_plugin i_s_innodb_table_stats_maria; +extern struct st_maria_plugin i_s_innodb_index_stats_maria; + #endif /* i_s_h */

1 0

Re: [Maria-developers] options for CREATE TABLE (MWL#43)
by Oleksandr Byelkin 11 Mar '10

11 Mar '10

Hi! Quick answer to solve "big" questions faster. 11 марта 2010, в 12:30, Sergei Golubchik написал(а): > Hi, Sanja! > > Here's the review, below: > > Summary: > > 1. please, store options together with the objects they describe, not > separately. .frm file is not the place I can stuck where I want, IMHO one extension as it stored now only the way to keep .frm compatible. > 2. Unknown option should be an error by default. > 3. use something my_getopt-like as we discussed, don't force every > engine to parse its options Above is exactly against Monty's expectations (I remember old discussion, as far as I remember my_getopt-like idea was rejected at the end, about error messages it can be done for creation, but for alter table we agreed to have warnings). [skip] > 5. don't check for changed options in alter table with your > check_if_incompatible_data. let the engine do that. Do you mean additional call to engine? [skip] > 7. parser: make the equal sign optional > 8. few existing options, like row_format, insert_method, checksum, > delay_key_write, key_block_size, min_rows/max_rows, avg_row_length, > tablespace, connection, pack_keys could be moved into storage > engines > out of the parser. In some cases options goes without coma, there is 3 word option DATA DIRECTORY <value>, INDEX DIRECTORY <value> so I can't imagine how to move the existing options to engine and make equal sign optional. [skip]

1 0

[Maria-developers] LIMIT optimisations
by Jocelyn Fournier 11 Mar '10

11 Mar '10

Hi, Following this discussion in 2007 : http://lists.mysql.com/internals/34287 Is there any plan to implement such an optimisation in MariaDB ? (I think a lot of web app using pagination could take benefit of such an optimisation, although there are some workarounds to avoid big LIMIT for pagination) Thanks ! Jocelyn

3 4

[Maria-developers] Need help with packaging for MariaDB 5.2
by Kristian Nielsen 10 Mar '10

10 Mar '10

Hi Arjen, We are starting to think about MariaDB 5.2, in particular of start making (alpha) releases with packages. The current ourdelta packaging scripts for MariaDB 5.1 fail to build 5.2 (seen in Buildbot). I took a look at it, unfortunately it is more complicated than what I can easily fix with my limited knowledge of packaging stuff. For .deb, the "5.1" is part of packaging names, which again is part of package dependencies. So it needs to be updated with correct names and "Replaces:", "Conflicts:", etc headers and so on. For .rpm the issue may be similar, not sure. Can you (or other from OurDelta) help me with this, or suggest the best way forward? - Kristian.

2 1

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport optimizations for derived tables and views (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: Server-9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Title modified. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -Backport optimizations for derived tables and views. +Backport optimizations for derived tables and views -=-=(Igor - Wed, 10 Mar 2010, 22:17)=-=- Version updated. --- /tmp/wklog.106.old.2763 2010-03-10 22:17:28.000000000 +0000 +++ /tmp/wklog.106.new.2763 2010-03-10 22:17:28.000000000 +0000 @@ -1 +1 @@ -WorkLog-3.4 +Server-9.x DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] New (by Igor): Backport optimizations for derived tables and views. (106)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport optimizations for derived tables and views. CREATION DATE..: Wed, 10 Mar 2010, 22:16 SUPERVISOR.....: Monty IMPLEMENTOR....: Igor COPIES TO......: Igor, Monty, Psergey, Sanja, Sergei, Timour CATEGORY.......: Server-Sprint TASK ID........: 106 (http://askmonty.org/worklog/?tid=106) VERSION........: WorkLog-3.4 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: DESCRIPTION: The goal of this task is to backport the implementation of the late materialization of derived tables and views and the additional optimizations for derived tables/views from MySQL 6.0 code line to MariaDB 5.3. Numerous bugs in the existing code concerning this functionality are to be fixed within this task. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:02)=-=- High Level Description modified. --- /tmp/wklog.90.old.2007 2010-03-10 22:02:23.000000000 +0000 +++ /tmp/wklog.90.new.2007 2010-03-10 22:02:23.000000000 +0000 @@ -13,8 +13,8 @@ for each record R2 in big_table such that oe=R1 pass R2 to output -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. +Semi-join materialization supports the inside-out strategy. This WL entry is +about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports the inside-out strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:02)=-=- High Level Description modified. --- /tmp/wklog.90.old.2007 2010-03-10 22:02:23.000000000 +0000 +++ /tmp/wklog.90.new.2007 2010-03-10 22:02:23.000000000 +0000 @@ -13,8 +13,8 @@ for each record R2 in big_table such that oe=R1 pass R2 to output -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. +Semi-join materialization supports the inside-out strategy. This WL entry is +about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports the inside-out strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:02)=-=- High Level Description modified. --- /tmp/wklog.90.old.2007 2010-03-10 22:02:23.000000000 +0000 +++ /tmp/wklog.90.new.2007 2010-03-10 22:02:23.000000000 +0000 @@ -13,8 +13,8 @@ for each record R2 in big_table such that oe=R1 pass R2 to output -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. +Semi-join materialization supports the inside-out strategy. This WL entry is +about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports the inside-out strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 22:02)=-=- High Level Description modified. --- /tmp/wklog.90.old.2007 2010-03-10 22:02:23.000000000 +0000 +++ /tmp/wklog.90.new.2007 2010-03-10 22:02:23.000000000 +0000 @@ -13,8 +13,8 @@ for each record R2 in big_table such that oe=R1 pass R2 to output -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. +Semi-join materialization supports the inside-out strategy. This WL entry is +about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports the inside-out strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE (90)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE CREATION DATE..: Sun, 28 Feb 2010, 13:45 SUPERVISOR.....: Monty IMPLEMENTOR....: Psergey COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 90 (http://askmonty.org/worklog/?tid=90) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: -1 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:52)=-=- Status updated. --- /tmp/wklog.90.old.882 2010-03-10 21:52:02.000000000 +0000 +++ /tmp/wklog.90.new.882 2010-03-10 21:52:02.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 15:37)=-=- High Level Description modified. --- /tmp/wklog.90.old.23524 2010-02-28 15:37:47.000000000 +0000 +++ /tmp/wklog.90.new.23524 2010-02-28 15:37:47.000000000 +0000 @@ -15,3 +15,7 @@ Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. + + +Once WL#89 is done, there will be a cost-based choice between +Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. -=-=(Psergey - Sun, 28 Feb 2010, 15:22)=-=- High-Level Specification modified. --- /tmp/wklog.90.old.23033 2010-02-28 15:22:09.000000000 +0000 +++ /tmp/wklog.90.new.23033 2010-02-28 15:22:09.000000000 +0000 @@ -1 +1,33 @@ +Basic idea on how this could be achieved: + +Pre-optimization phase +---------------------- + +The rewrite +~~~~~~~~~~~ +If we find a subquery predicate that is +- not processed by current semi-join optimizations +- is an AND-part of the WHERE/ON clause +- can be executed with Materialization + +then +- Remove the predicate from WHERE/ON clause +- Add a special JOIN_TAB object instead. + +Plan options +~~~~~~~~~~~~ +- Use the IN-equality to create KEYUSE elements. + +Optimization +------------ +- Pre-optimize the subquery so we know materialization cost +- Whenever best_access_path() encounters the "special JOIN_TAB" it should + consider two strategies: + A. Materialization and making lookups in the materialized table (if applicable) + B. Materialization and then scanning the materialized table. + + +EXPLAIN +------- +TODO how this will look in EXPLAIN output? -=-=(Psergey - Sun, 28 Feb 2010, 14:56)=-=- Dependency created: 91 now depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:54)=-=- Dependency deleted: 94 no longer depends on 90 -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21903 2010-02-28 14:47:54.000000000 +0000 +++ /tmp/wklog.90.new.21903 2010-02-28 14:47:54.000000000 +0000 @@ -1 +1 @@ - Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- High Level Description modified. --- /tmp/wklog.90.old.21880 2010-02-28 14:47:28.000000000 +0000 +++ /tmp/wklog.90.new.21880 2010-02-28 14:47:28.000000000 +0000 @@ -1,10 +1,17 @@ -For uncorrelated IN subqueries that can't be converted to semi-joins it is -necessary to make a cost-based choice between IN->EXISTS and Materialization -strategies. +Consider the following case: -Both strategies handle two cases: -1. A simple case w/o NULLs handling -2. Handling NULLs. +SELECT * FROM big_table +WHERE oe IN (SELECT ie FROM table_with_few_groups + WHERE ... + GROUP BY group_col) AND ... -This WL is about making cost-based decision for #1. +Here the best way to execute the query is: + Materialize the subquery; + # now run the join: + for each record R1 in materialized table + for each record R2 in big_table such that oe=R1 + pass R2 to output + +Semi-join materialization supports such strategy with SJM-Scan strategy. This WL +entry is about adding support for such strategies for non-semijoin subqueries. -=-=(Psergey - Sun, 28 Feb 2010, 14:47)=-=- Title modified. --- /tmp/wklog.90.old.21859 2010-02-28 14:47:02.000000000 +0000 +++ /tmp/wklog.90.new.21859 2010-02-28 14:47:02.000000000 +0000 @@ -1 +1 @@ -Subqueries: cost-based choice between Materialization and IN->EXISTS transformation + Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE -=-=(Psergey - Sun, 28 Feb 2010, 14:08)=-=- Dependency created: 94 now depends on 90 DESCRIPTION: Consider the following case: SELECT * FROM big_table WHERE oe IN (SELECT ie FROM table_with_few_groups WHERE ... GROUP BY group_col) AND ... Here the best way to execute the query is: Materialize the subquery; # now run the join: for each record R1 in materialized table for each record R2 in big_table such that oe=R1 pass R2 to output Semi-join materialization supports such strategy with SJM-Scan strategy. This WL entry is about adding support for such strategies for non-semijoin subqueries. Once WL#89 is done, there will be a cost-based choice between Materialization+lookup, Materialization+scan, and IN->EXISTS+lookup strategies. HIGH-LEVEL SPECIFICATION: Basic idea on how this could be achieved: Pre-optimization phase ---------------------- The rewrite ~~~~~~~~~~~ If we find a subquery predicate that is - not processed by current semi-join optimizations - is an AND-part of the WHERE/ON clause - can be executed with Materialization then - Remove the predicate from WHERE/ON clause - Add a special JOIN_TAB object instead. Plan options ~~~~~~~~~~~~ - Use the IN-equality to create KEYUSE elements. Optimization ------------ - Pre-optimize the subquery so we know materialization cost - Whenever best_access_path() encounters the "special JOIN_TAB" it should consider two strategies: A. Materialization and making lookups in the materialized table (if applicable) B. Materialization and then scanning the materialized table. EXPLAIN ------- TODO how this will look in EXPLAIN output? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries: cost-based choice between Materialization and IN->EXISTS transformation (89)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries: cost-based choice between Materialization and IN->EXISTS transformation CREATION DATE..: Sun, 28 Feb 2010, 13:39 SUPERVISOR.....: Monty IMPLEMENTOR....: Timour COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-Sprint TASK ID........: 89 (http://askmonty.org/worklog/?tid=89) VERSION........: Server-5.3 STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Category updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:48)=-=- Status updated. --- /tmp/wklog.89.old.778 2010-03-10 21:48:08.000000000 +0000 +++ /tmp/wklog.89.new.778 2010-03-10 21:48:08.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Assigned -=-=(Psergey - Sun, 28 Feb 2010, 16:34)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24497 2010-02-28 16:34:05.000000000 +0000 +++ /tmp/wklog.89.new.24497 2010-02-28 16:34:05.000000000 +0000 @@ -36,8 +36,8 @@ So, we'll need to compute both exists_select_cost and materialization_cost. -Difficulty with computing the two costs ---------------------------------------- +Difficulty with the need to run select optimization two times +------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. @@ -46,4 +46,10 @@ 3. Then we find that cost #1 is less and want to execute the materialization strategy. +The problem is that once one injects "oe=ie", it can trigger some optimization +steps that are not possible to undo. +- Example1: outer->inner join conversion +- non-Example: according to Igor, "oe=ie" won't participate in equality propagation. +- ... what else ? + -=-=(Psergey - Sun, 28 Feb 2010, 16:08)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24098 2010-02-28 16:08:56.000000000 +0000 +++ /tmp/wklog.89.new.24098 2010-02-28 16:08:56.000000000 +0000 @@ -36,3 +36,14 @@ So, we'll need to compute both exists_select_cost and materialization_cost. +Difficulty with computing the two costs +--------------------------------------- +The problem is in this scenario: +1. We compute materialization_cost by running optimization for the original + subquery select. +2. We compute exists_select_cost by running optimization for the subquery's + select with "oe=ie" injected into WHERE +3. Then we find that cost #1 is less and want to execute the materialization + strategy. + + -=-=(Psergey - Sun, 28 Feb 2010, 15:57)=-=- High-Level Specification modified. --- /tmp/wklog.89.old.24045 2010-02-28 15:57:49.000000000 +0000 +++ /tmp/wklog.89.new.24045 2010-02-28 15:57:49.000000000 +0000 @@ -1 +1,38 @@ +Why need two optimizations +-------------------------- +Consider a query with subquery: + + SELECT + oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) + FROM outer_tbl + WHERE outer_cond + +If we use Materialization strategy, the costs will be + + cost of accessing outer_tbl + + materialization_cost + + #records(outer_tbl w/o outer_cond) * lookup_cost + +where + + materialization_cost= + cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) + +On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into + + SELECT + EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + FROM outer_tbl + WHERE outer_cond + +and the costs will be + + cost of accessing outer_tbl + + #records(outer_tbl w/o outer_cond) * exists_select_cost + +where + exists_select_cost= + cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) + +So, we'll need to compute both exists_select_cost and materialization_cost. -=-=(Psergey - Sun, 28 Feb 2010, 15:07)=-=- Dependency created: 91 now depends on 89 DESCRIPTION: For uncorrelated IN subqueries that can't be converted to semi-joins it is necessary to make a cost-based choice between IN->EXISTS and Materialization strategies. Both strategies handle two cases: 1. A simple case w/o NULLs handling 2. Handling NULLs. This WL is about making cost-based decision for #1. HIGH-LEVEL SPECIFICATION: Why need two optimizations -------------------------- Consider a query with subquery: SELECT oe IN (SELECT ie FROM inner_tbl WHERE inner_cond) FROM outer_tbl WHERE outer_cond If we use Materialization strategy, the costs will be cost of accessing outer_tbl + materialization_cost + #records(outer_tbl w/o outer_cond) * lookup_cost where materialization_cost= cost of executing the (SELECT ie FROM inner_tbl WHERE inner_cond) On the other hand, for IN->EXISTS strategy, the subquery will be rewritten into SELECT EXISTS (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) FROM outer_tbl WHERE outer_cond and the costs will be cost of accessing outer_tbl + #records(outer_tbl w/o outer_cond) * exists_select_cost where exists_select_cost= cost of executing the (SELECT 1 FROM inner_tbl WHERE inner_cond AND oe=ie) So, we'll need to compute both exists_select_cost and materialization_cost. Difficulty with the need to run select optimization two times ------------------------------------------------------------- The problem is in this scenario: 1. We compute materialization_cost by running optimization for the original subquery select. 2. We compute exists_select_cost by running optimization for the subquery's select with "oe=ie" injected into WHERE 3. Then we find that cost #1 is less and want to execute the materialization strategy. The problem is that once one injects "oe=ie", it can trigger some optimization steps that are not possible to undo. - Example1: outer->inner join conversion - non-Example: according to Igor, "oe=ie" won't participate in equality propagation. - ... what else ? ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries backport: fix known semi-join subquery bugs (92)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries backport: fix known semi-join subquery bugs CREATION DATE..: Sun, 28 Feb 2010, 14:02 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 92 (http://askmonty.org/worklog/?tid=92) VERSION........: WorkLog-3.4 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:33)=-=- High Level Description modified. --- /tmp/wklog.92.old.32291 2010-03-10 21:33:29.000000000 +0000 +++ /tmp/wklog.92.new.32291 2010-03-10 21:33:29.000000000 +0000 @@ -1,3 +1,5 @@ +The goal of this task is to fix all known subquery semi-join bugs. + We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. -=-=(Psergey - Sun, 28 Feb 2010, 16:41)=-=- High Level Description modified. --- /tmp/wklog.92.old.24539 2010-02-28 16:41:06.000000000 +0000 +++ /tmp/wklog.92.new.24539 2010-02-28 16:41:06.000000000 +0000 @@ -1 +1,4 @@ We must fix known subquery semi-join bugs. +* outer join + semi join problem +* Duplicate Weedout + join caching problem. + -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- High Level Description modified. --- /tmp/wklog.92.old.22593 2010-02-28 15:06:23.000000000 +0000 +++ /tmp/wklog.92.new.22593 2010-02-28 15:06:23.000000000 +0000 @@ -1 +1 @@ - +We must fix known subquery semi-join bugs. -=-=(Psergey - Sun, 28 Feb 2010, 15:03)=-=- Title modified. --- /tmp/wklog.92.old.22572 2010-02-28 15:03:51.000000000 +0000 +++ /tmp/wklog.92.new.22572 2010-02-28 15:03:51.000000000 +0000 @@ -1 +1 @@ -Unused +Subqueries backport: fix known semi-join subquery bugs -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency deleted: 91 no longer depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- High Level Description modified. --- /tmp/wklog.92.old.22267 2010-02-28 14:57:52.000000000 +0000 +++ /tmp/wklog.92.new.22267 2010-02-28 14:57:52.000000000 +0000 @@ -1 +1 @@ -We must fix known semi-join subquery bugs. + -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- Title modified. --- /tmp/wklog.92.old.22249 2010-02-28 14:57:41.000000000 +0000 +++ /tmp/wklog.92.new.22249 2010-02-28 14:57:41.000000000 +0000 @@ -1 +1 @@ -Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Unused -=-=(Psergey - Sun, 28 Feb 2010, 14:51)=-=- High Level Description modified. --- /tmp/wklog.92.old.21961 2010-02-28 14:51:06.000000000 +0000 +++ /tmp/wklog.92.new.21961 2010-02-28 14:51:06.000000000 +0000 @@ -1,18 +1 @@ -Consider the following case: - -SELECT * FROM big_table -WHERE oe IN (SELECT ie FROM table_with_few_groups - WHERE ... - GROUP BY group_col) AND ... - -Here the best way to execute the query is: - - Materialize the subquery; - # now run the join: - for each record R1 in materialized table - for each record R2 in big_table such that oe=R1 - pass R2 to output - -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. - +We must fix known semi-join subquery bugs. ------------------------------------------------------------ -=-=(View All Progress Notes, 11 total)=-=- http://askmonty.org/worklog/index.pl?tid=92&nolimit=1 DESCRIPTION: The goal of this task is to fix all known subquery semi-join bugs. We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries backport: fix known semi-join subquery bugs (92)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries backport: fix known semi-join subquery bugs CREATION DATE..: Sun, 28 Feb 2010, 14:02 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 92 (http://askmonty.org/worklog/?tid=92) VERSION........: WorkLog-3.4 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:33)=-=- High Level Description modified. --- /tmp/wklog.92.old.32291 2010-03-10 21:33:29.000000000 +0000 +++ /tmp/wklog.92.new.32291 2010-03-10 21:33:29.000000000 +0000 @@ -1,3 +1,5 @@ +The goal of this task is to fix all known subquery semi-join bugs. + We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. -=-=(Psergey - Sun, 28 Feb 2010, 16:41)=-=- High Level Description modified. --- /tmp/wklog.92.old.24539 2010-02-28 16:41:06.000000000 +0000 +++ /tmp/wklog.92.new.24539 2010-02-28 16:41:06.000000000 +0000 @@ -1 +1,4 @@ We must fix known subquery semi-join bugs. +* outer join + semi join problem +* Duplicate Weedout + join caching problem. + -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- High Level Description modified. --- /tmp/wklog.92.old.22593 2010-02-28 15:06:23.000000000 +0000 +++ /tmp/wklog.92.new.22593 2010-02-28 15:06:23.000000000 +0000 @@ -1 +1 @@ - +We must fix known subquery semi-join bugs. -=-=(Psergey - Sun, 28 Feb 2010, 15:03)=-=- Title modified. --- /tmp/wklog.92.old.22572 2010-02-28 15:03:51.000000000 +0000 +++ /tmp/wklog.92.new.22572 2010-02-28 15:03:51.000000000 +0000 @@ -1 +1 @@ -Unused +Subqueries backport: fix known semi-join subquery bugs -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency deleted: 91 no longer depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- High Level Description modified. --- /tmp/wklog.92.old.22267 2010-02-28 14:57:52.000000000 +0000 +++ /tmp/wklog.92.new.22267 2010-02-28 14:57:52.000000000 +0000 @@ -1 +1 @@ -We must fix known semi-join subquery bugs. + -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- Title modified. --- /tmp/wklog.92.old.22249 2010-02-28 14:57:41.000000000 +0000 +++ /tmp/wklog.92.new.22249 2010-02-28 14:57:41.000000000 +0000 @@ -1 +1 @@ -Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Unused -=-=(Psergey - Sun, 28 Feb 2010, 14:51)=-=- High Level Description modified. --- /tmp/wklog.92.old.21961 2010-02-28 14:51:06.000000000 +0000 +++ /tmp/wklog.92.new.21961 2010-02-28 14:51:06.000000000 +0000 @@ -1,18 +1 @@ -Consider the following case: - -SELECT * FROM big_table -WHERE oe IN (SELECT ie FROM table_with_few_groups - WHERE ... - GROUP BY group_col) AND ... - -Here the best way to execute the query is: - - Materialize the subquery; - # now run the join: - for each record R1 in materialized table - for each record R2 in big_table such that oe=R1 - pass R2 to output - -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. - +We must fix known semi-join subquery bugs. ------------------------------------------------------------ -=-=(View All Progress Notes, 11 total)=-=- http://askmonty.org/worklog/index.pl?tid=92&nolimit=1 DESCRIPTION: The goal of this task is to fix all known subquery semi-join bugs. We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries backport: fix known semi-join subquery bugs (92)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries backport: fix known semi-join subquery bugs CREATION DATE..: Sun, 28 Feb 2010, 14:02 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 92 (http://askmonty.org/worklog/?tid=92) VERSION........: WorkLog-3.4 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:33)=-=- High Level Description modified. --- /tmp/wklog.92.old.32291 2010-03-10 21:33:29.000000000 +0000 +++ /tmp/wklog.92.new.32291 2010-03-10 21:33:29.000000000 +0000 @@ -1,3 +1,5 @@ +The goal of this task is to fix all known subquery semi-join bugs. + We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. -=-=(Psergey - Sun, 28 Feb 2010, 16:41)=-=- High Level Description modified. --- /tmp/wklog.92.old.24539 2010-02-28 16:41:06.000000000 +0000 +++ /tmp/wklog.92.new.24539 2010-02-28 16:41:06.000000000 +0000 @@ -1 +1,4 @@ We must fix known subquery semi-join bugs. +* outer join + semi join problem +* Duplicate Weedout + join caching problem. + -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- High Level Description modified. --- /tmp/wklog.92.old.22593 2010-02-28 15:06:23.000000000 +0000 +++ /tmp/wklog.92.new.22593 2010-02-28 15:06:23.000000000 +0000 @@ -1 +1 @@ - +We must fix known subquery semi-join bugs. -=-=(Psergey - Sun, 28 Feb 2010, 15:03)=-=- Title modified. --- /tmp/wklog.92.old.22572 2010-02-28 15:03:51.000000000 +0000 +++ /tmp/wklog.92.new.22572 2010-02-28 15:03:51.000000000 +0000 @@ -1 +1 @@ -Unused +Subqueries backport: fix known semi-join subquery bugs -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency deleted: 91 no longer depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- High Level Description modified. --- /tmp/wklog.92.old.22267 2010-02-28 14:57:52.000000000 +0000 +++ /tmp/wklog.92.new.22267 2010-02-28 14:57:52.000000000 +0000 @@ -1 +1 @@ -We must fix known semi-join subquery bugs. + -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- Title modified. --- /tmp/wklog.92.old.22249 2010-02-28 14:57:41.000000000 +0000 +++ /tmp/wklog.92.new.22249 2010-02-28 14:57:41.000000000 +0000 @@ -1 +1 @@ -Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Unused -=-=(Psergey - Sun, 28 Feb 2010, 14:51)=-=- High Level Description modified. --- /tmp/wklog.92.old.21961 2010-02-28 14:51:06.000000000 +0000 +++ /tmp/wklog.92.new.21961 2010-02-28 14:51:06.000000000 +0000 @@ -1,18 +1 @@ -Consider the following case: - -SELECT * FROM big_table -WHERE oe IN (SELECT ie FROM table_with_few_groups - WHERE ... - GROUP BY group_col) AND ... - -Here the best way to execute the query is: - - Materialize the subquery; - # now run the join: - for each record R1 in materialized table - for each record R2 in big_table such that oe=R1 - pass R2 to output - -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. - +We must fix known semi-join subquery bugs. ------------------------------------------------------------ -=-=(View All Progress Notes, 11 total)=-=- http://askmonty.org/worklog/index.pl?tid=92&nolimit=1 DESCRIPTION: The goal of this task is to fix all known subquery semi-join bugs. We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subqueries backport: fix known semi-join subquery bugs (92)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subqueries backport: fix known semi-join subquery bugs CREATION DATE..: Sun, 28 Feb 2010, 14:02 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Igor, Psergey, Timour CATEGORY.......: Server-RawIdeaBin TASK ID........: 92 (http://askmonty.org/worklog/?tid=92) VERSION........: WorkLog-3.4 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:33)=-=- High Level Description modified. --- /tmp/wklog.92.old.32291 2010-03-10 21:33:29.000000000 +0000 +++ /tmp/wklog.92.new.32291 2010-03-10 21:33:29.000000000 +0000 @@ -1,3 +1,5 @@ +The goal of this task is to fix all known subquery semi-join bugs. + We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. -=-=(Psergey - Sun, 28 Feb 2010, 16:41)=-=- High Level Description modified. --- /tmp/wklog.92.old.24539 2010-02-28 16:41:06.000000000 +0000 +++ /tmp/wklog.92.new.24539 2010-02-28 16:41:06.000000000 +0000 @@ -1 +1,4 @@ We must fix known subquery semi-join bugs. +* outer join + semi join problem +* Duplicate Weedout + join caching problem. + -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 15:06)=-=- High Level Description modified. --- /tmp/wklog.92.old.22593 2010-02-28 15:06:23.000000000 +0000 +++ /tmp/wklog.92.new.22593 2010-02-28 15:06:23.000000000 +0000 @@ -1 +1 @@ - +We must fix known subquery semi-join bugs. -=-=(Psergey - Sun, 28 Feb 2010, 15:03)=-=- Title modified. --- /tmp/wklog.92.old.22572 2010-02-28 15:03:51.000000000 +0000 +++ /tmp/wklog.92.new.22572 2010-02-28 15:03:51.000000000 +0000 @@ -1 +1 @@ -Unused +Subqueries backport: fix known semi-join subquery bugs -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency deleted: 91 no longer depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:58)=-=- Dependency created: 91 now depends on 92 -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- High Level Description modified. --- /tmp/wklog.92.old.22267 2010-02-28 14:57:52.000000000 +0000 +++ /tmp/wklog.92.new.22267 2010-02-28 14:57:52.000000000 +0000 @@ -1 +1 @@ -We must fix known semi-join subquery bugs. + -=-=(Psergey - Sun, 28 Feb 2010, 14:57)=-=- Title modified. --- /tmp/wklog.92.old.22249 2010-02-28 14:57:41.000000000 +0000 +++ /tmp/wklog.92.new.22249 2010-02-28 14:57:41.000000000 +0000 @@ -1 +1 @@ -Subqueries: Inside-out execution for non-semijoin materialized subqueries that are AND-parts of the WHERE +Unused -=-=(Psergey - Sun, 28 Feb 2010, 14:51)=-=- High Level Description modified. --- /tmp/wklog.92.old.21961 2010-02-28 14:51:06.000000000 +0000 +++ /tmp/wklog.92.new.21961 2010-02-28 14:51:06.000000000 +0000 @@ -1,18 +1 @@ -Consider the following case: - -SELECT * FROM big_table -WHERE oe IN (SELECT ie FROM table_with_few_groups - WHERE ... - GROUP BY group_col) AND ... - -Here the best way to execute the query is: - - Materialize the subquery; - # now run the join: - for each record R1 in materialized table - for each record R2 in big_table such that oe=R1 - pass R2 to output - -Semi-join materialization supports such strategy with SJM-Scan strategy. This WL -entry is about adding support for such strategies for non-semijoin subqueries. - +We must fix known semi-join subquery bugs. ------------------------------------------------------------ -=-=(View All Progress Notes, 11 total)=-=- http://askmonty.org/worklog/index.pl?tid=92&nolimit=1 DESCRIPTION: The goal of this task is to fix all known subquery semi-join bugs. We must fix known subquery semi-join bugs. * outer join + semi join problem * Duplicate Weedout + join caching problem. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache (66)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache CREATION DATE..: Wed, 25 Nov 2009, 22:25 SUPERVISOR.....: Monty IMPLEMENTOR....: Sanja COPIES TO......: CATEGORY.......: Client-Sprint TASK ID........: 66 (http://askmonty.org/worklog/?tid=66) VERSION........: 9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:29)=-=- High Level Description modified. --- /tmp/wklog.66.old.32188 2010-03-10 21:29:16.000000000 +0000 +++ /tmp/wklog.66.new.32188 2010-03-10 21:29:16.000000000 +0000 @@ -1,3 +1,10 @@ +The goal of this task is to optimize evaluation of subqueries and subquery +predicates by storing the results of a correlated subquery together with +correlation parameters in a cache and reusing those results for the same sets of +parameters. + +Here's what is to be done in this task in more details: + Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and -=-=(Igor - Wed, 10 Mar 2010, 21:13)=-=- Dependency created: 91 now depends on 66 -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Category updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Client-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Version updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-5.3 +9.x -=-=(Monty - Fri, 29 Jan 2010, 19:07)=-=- Version updated. --- /tmp/wklog.66.old.5893 2010-01-29 19:07:10.000000000 +0200 +++ /tmp/wklog.66.new.5893 2010-01-29 19:07:10.000000000 +0200 @@ -1 +1 @@ -Server-5.2 +Server-5.3 -=-=(Psergey - Wed, 20 Jan 2010, 14:50)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.26873 2010-01-20 14:50:41.000000000 +0200 +++ /tmp/wklog.66.new.26873 2010-01-20 14:50:41.000000000 +0200 @@ -4,7 +4,6 @@ To check/discuss: ----------------- -* Do we put subquery cache on all levels of subqueries or on highest level only * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* @@ -41,7 +40,12 @@ - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. -We dont support cases when outer_expr or correlation_references are blobs. +We don't support cases when outer_expr or correlation_references are blobs. + +All subquery predicates are cached. That is, if one subquery predicate is +located within another, both of them will have caches (one option to reduce +cache memory usage was to use cache only for the upper-most select. we decided +against it). 2. Data structure used for the cache ------------------------------------ -=-=(Psergey - Wed, 20 Jan 2010, 13:07)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.17649 2010-01-20 13:07:07.000000000 +0200 +++ /tmp/wklog.66.new.17649 2010-01-20 13:07:07.000000000 +0200 @@ -3,7 +3,13 @@ To check/discuss: - To put subquery cache on all levels of subqueries or on highest level only. +----------------- +* Do we put subquery cache on all levels of subqueries or on highest level only +* Will there be any means to measure subquery cache hit rate? +* MySQL-6.0 has a one-element predicate result cache. It is called "left + expression cache", grep for left_expr_cache in sql/item_subselect.* + When this WL is merged with 6.0's optimizations, these two caches will + need to be unified somehow. <contents> -=-=(Psergey - Mon, 18 Jan 2010, 16:40)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24899 2010-01-18 16:40:16.000000000 +0200 +++ /tmp/wklog.66.new.24899 2010-01-18 16:40:16.000000000 +0200 @@ -1,3 +1,5 @@ +* Target version: base on mysql-5.2 code + All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) -=-=(Psergey - Mon, 18 Jan 2010, 16:37)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24586 2010-01-18 16:37:07.000000000 +0200 +++ /tmp/wklog.66.new.24586 2010-01-18 16:37:07.000000000 +0200 @@ -4,6 +4,11 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to fill the temptable +------------------------- +Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() +and is supposed to be quite trivial. + How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. -=-=(Psergey - Mon, 18 Jan 2010, 16:34)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24328 2010-01-18 16:34:19.000000000 +0200 +++ /tmp/wklog.66.new.24328 2010-01-18 16:34:19.000000000 +0200 @@ -32,8 +32,8 @@ Question: or perhaps that is not necessarry? </questionable> -Execution process -~~~~~~~~~~~~~~~~~ +Doing the lookup +~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ @@ -42,4 +42,12 @@ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); +The code in this WL will use the same approach +Extracting the value of the subquery predicate +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The goal of making the lookup is to get the value of subquery predicate. +This is done by creating an Item_field $I which refers to appropriate +temporary table's field and then subquery_predicate->val_int() will invoke +$I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so +forth. ------------------------------------------------------------ -=-=(View All Progress Notes, 17 total)=-=- http://askmonty.org/worklog/index.pl?tid=66&nolimit=1 DESCRIPTION: The goal of this task is to optimize evaluation of subqueries and subquery predicates by storing the results of a correlated subquery together with correlation parameters in a cache and reusing those results for the same sets of parameters. Here's what is to be done in this task in more details: Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and return cached result of subquery if the reference values combination has already been used. For example in the following subquery: (L1, L2) IN (SELECT A, B FROM T WHERE T.F1>OTER_FIELD) set of references to look into the subquery cache is (L1, L2, OTER_FIELD). The subquery cache should be implemented as simple LRU connected to the subquery. Size of the subquery cache (in number of results (but maybe in used memory amount)) is limited by session variable (query parameter?). HIGH-LEVEL SPECIFICATION: Attach subquery cache to each Item_subquery. Interface should allow to use hash or temporary table inside. To check/discuss: ----------------- * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* When this WL is merged with 6.0's optimizations, these two caches will need to be unified somehow. <contents> 1. Scope of the task 2. Data structure used for the cache 3. Cache size 4. Interplay with other subquery optimizations 5. User interface </contents> 1. Scope of the task -------------------- This WL should handle all subquery predicates, i.e. it should handle these cases: outer_expr IN (SELECT correlated_select) outer_expr $CMP$ ALL/ANY (SELECT correlated_select) EXISTS (SELECT correlated_select) scalar-context subquery: (SELECT correlated_select) The cache will maintain (outer_expr, correlation_references)-> subquery_item_result mapping, where - correlation_references is a list of tablename.column_name that are referred from the correlated_select but tablename is a table that is ouside the subquery. - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. We don't support cases when outer_expr or correlation_references are blobs. All subquery predicates are cached. That is, if one subquery predicate is located within another, both of them will have caches (one option to reduce cache memory usage was to use cache only for the upper-most select. we decided against it). 2. Data structure used for the cache ------------------------------------ There are two data structures available in the codebase that will allow fast equality lookups: 1. HASH (mysys/hash.c) tables 2. Temporary tables (the ones that are used for e.g. GROUP BY) None of them has any support for element eviction on overflow (using LRU or some other policy). Query cache and MyISAM/Maria's key/page cache ought to support some eviction mechanism, but code-wise it is not readily reusable, one will need to factor it out (or copy it). We choose to use #2, and not to have any eviction policy. See subsequent sections for details and reasoning behind the decision. 3. Cache size ------------- Typically, a cache has some maximum size and a policy which is used to select a cache entry for removal when the cache becomes full (e.g. find and remove the least [recently] used entry) For this WL entry we will use a cache of infinite size. The reasoning behind this is that: - is is easy to do: we have temporary tables that can grow to arbitrarily large size while still providing the same insert/lookup interface. - it suits us: unless the subquery is resolved with one index lookup, hitting the cache would be many times cheaper than re-running the subquery, so cache is worth having. 4. Interplay with other subquery optimizations ---------------------------------------------- * This WL entry should not care about IN->EXISTS transformation: caching for IN subquery and result of its conversion to EXISTS would work in the same way. * This optimization is orthogonal to <=>ANY -> MIN/MAX rewrite (it will work/be useful irrespectively of whether the rewrite has been performed or not) * TODO: compare this with materialization for uncorrelated IN-subqueries. Is this basically the same? A: no, it is not: - IN-Materialization has to perform full materialization before it can do the first subquery evaluation. This WL's code has almost no startup costs. - This optimization has temp.table of (corr_reference, predicate_value), while IN-materialization will have (corr_reference) only. 5. User interface ----------------- * There will be an @@optimizer_switch flag to turn this optimization on and off (TODO: name of the flag?) * TODO: how do we show this in EXPLAIN [EXTENDED]? The most easiest is to print something in the warning text of EXPLAIN EXTEDED that would indicate use of cache. * temporary table sizing (max size for heap table, whether to use MyISAM or Maria) will be controlled with common temp.table control variables. LOW-LEVEL DESIGN: * Target version: base on mysql-5.2 code All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). How to fill the temptable ------------------------- Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() and is supposed to be quite trivial. How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. Setup process ~~~~~~~~~~~~~ Setup is performed in the same way as in setup_sj_materialization(), see the code that starts these lines: /* Create/initialize everything we will need to index lookups into the temptable. */ and ends at this line: Remove the injected semi-join IN-equalities from join_tab conds. This <questionable> We'll also need to check equalities, i.e. do an equivalent of this: if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, emb_sj_nest->sj_subq_pred))) DBUG_RETURN(TRUE); /* purecov: inspected */ Question: or perhaps that is not necessarry? </questionable> Doing the lookup ~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); The code in this WL will use the same approach Extracting the value of the subquery predicate ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The goal of making the lookup is to get the value of subquery predicate. This is done by creating an Item_field $I which refers to appropriate temporary table's field and then subquery_predicate->val_int() will invoke $I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so forth. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache (66)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache CREATION DATE..: Wed, 25 Nov 2009, 22:25 SUPERVISOR.....: Monty IMPLEMENTOR....: Sanja COPIES TO......: CATEGORY.......: Client-Sprint TASK ID........: 66 (http://askmonty.org/worklog/?tid=66) VERSION........: 9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:29)=-=- High Level Description modified. --- /tmp/wklog.66.old.32188 2010-03-10 21:29:16.000000000 +0000 +++ /tmp/wklog.66.new.32188 2010-03-10 21:29:16.000000000 +0000 @@ -1,3 +1,10 @@ +The goal of this task is to optimize evaluation of subqueries and subquery +predicates by storing the results of a correlated subquery together with +correlation parameters in a cache and reusing those results for the same sets of +parameters. + +Here's what is to be done in this task in more details: + Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and -=-=(Igor - Wed, 10 Mar 2010, 21:13)=-=- Dependency created: 91 now depends on 66 -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Category updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Client-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Version updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-5.3 +9.x -=-=(Monty - Fri, 29 Jan 2010, 19:07)=-=- Version updated. --- /tmp/wklog.66.old.5893 2010-01-29 19:07:10.000000000 +0200 +++ /tmp/wklog.66.new.5893 2010-01-29 19:07:10.000000000 +0200 @@ -1 +1 @@ -Server-5.2 +Server-5.3 -=-=(Psergey - Wed, 20 Jan 2010, 14:50)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.26873 2010-01-20 14:50:41.000000000 +0200 +++ /tmp/wklog.66.new.26873 2010-01-20 14:50:41.000000000 +0200 @@ -4,7 +4,6 @@ To check/discuss: ----------------- -* Do we put subquery cache on all levels of subqueries or on highest level only * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* @@ -41,7 +40,12 @@ - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. -We dont support cases when outer_expr or correlation_references are blobs. +We don't support cases when outer_expr or correlation_references are blobs. + +All subquery predicates are cached. That is, if one subquery predicate is +located within another, both of them will have caches (one option to reduce +cache memory usage was to use cache only for the upper-most select. we decided +against it). 2. Data structure used for the cache ------------------------------------ -=-=(Psergey - Wed, 20 Jan 2010, 13:07)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.17649 2010-01-20 13:07:07.000000000 +0200 +++ /tmp/wklog.66.new.17649 2010-01-20 13:07:07.000000000 +0200 @@ -3,7 +3,13 @@ To check/discuss: - To put subquery cache on all levels of subqueries or on highest level only. +----------------- +* Do we put subquery cache on all levels of subqueries or on highest level only +* Will there be any means to measure subquery cache hit rate? +* MySQL-6.0 has a one-element predicate result cache. It is called "left + expression cache", grep for left_expr_cache in sql/item_subselect.* + When this WL is merged with 6.0's optimizations, these two caches will + need to be unified somehow. <contents> -=-=(Psergey - Mon, 18 Jan 2010, 16:40)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24899 2010-01-18 16:40:16.000000000 +0200 +++ /tmp/wklog.66.new.24899 2010-01-18 16:40:16.000000000 +0200 @@ -1,3 +1,5 @@ +* Target version: base on mysql-5.2 code + All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) -=-=(Psergey - Mon, 18 Jan 2010, 16:37)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24586 2010-01-18 16:37:07.000000000 +0200 +++ /tmp/wklog.66.new.24586 2010-01-18 16:37:07.000000000 +0200 @@ -4,6 +4,11 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to fill the temptable +------------------------- +Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() +and is supposed to be quite trivial. + How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. -=-=(Psergey - Mon, 18 Jan 2010, 16:34)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24328 2010-01-18 16:34:19.000000000 +0200 +++ /tmp/wklog.66.new.24328 2010-01-18 16:34:19.000000000 +0200 @@ -32,8 +32,8 @@ Question: or perhaps that is not necessarry? </questionable> -Execution process -~~~~~~~~~~~~~~~~~ +Doing the lookup +~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ @@ -42,4 +42,12 @@ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); +The code in this WL will use the same approach +Extracting the value of the subquery predicate +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The goal of making the lookup is to get the value of subquery predicate. +This is done by creating an Item_field $I which refers to appropriate +temporary table's field and then subquery_predicate->val_int() will invoke +$I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so +forth. ------------------------------------------------------------ -=-=(View All Progress Notes, 17 total)=-=- http://askmonty.org/worklog/index.pl?tid=66&nolimit=1 DESCRIPTION: The goal of this task is to optimize evaluation of subqueries and subquery predicates by storing the results of a correlated subquery together with correlation parameters in a cache and reusing those results for the same sets of parameters. Here's what is to be done in this task in more details: Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and return cached result of subquery if the reference values combination has already been used. For example in the following subquery: (L1, L2) IN (SELECT A, B FROM T WHERE T.F1>OTER_FIELD) set of references to look into the subquery cache is (L1, L2, OTER_FIELD). The subquery cache should be implemented as simple LRU connected to the subquery. Size of the subquery cache (in number of results (but maybe in used memory amount)) is limited by session variable (query parameter?). HIGH-LEVEL SPECIFICATION: Attach subquery cache to each Item_subquery. Interface should allow to use hash or temporary table inside. To check/discuss: ----------------- * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* When this WL is merged with 6.0's optimizations, these two caches will need to be unified somehow. <contents> 1. Scope of the task 2. Data structure used for the cache 3. Cache size 4. Interplay with other subquery optimizations 5. User interface </contents> 1. Scope of the task -------------------- This WL should handle all subquery predicates, i.e. it should handle these cases: outer_expr IN (SELECT correlated_select) outer_expr $CMP$ ALL/ANY (SELECT correlated_select) EXISTS (SELECT correlated_select) scalar-context subquery: (SELECT correlated_select) The cache will maintain (outer_expr, correlation_references)-> subquery_item_result mapping, where - correlation_references is a list of tablename.column_name that are referred from the correlated_select but tablename is a table that is ouside the subquery. - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. We don't support cases when outer_expr or correlation_references are blobs. All subquery predicates are cached. That is, if one subquery predicate is located within another, both of them will have caches (one option to reduce cache memory usage was to use cache only for the upper-most select. we decided against it). 2. Data structure used for the cache ------------------------------------ There are two data structures available in the codebase that will allow fast equality lookups: 1. HASH (mysys/hash.c) tables 2. Temporary tables (the ones that are used for e.g. GROUP BY) None of them has any support for element eviction on overflow (using LRU or some other policy). Query cache and MyISAM/Maria's key/page cache ought to support some eviction mechanism, but code-wise it is not readily reusable, one will need to factor it out (or copy it). We choose to use #2, and not to have any eviction policy. See subsequent sections for details and reasoning behind the decision. 3. Cache size ------------- Typically, a cache has some maximum size and a policy which is used to select a cache entry for removal when the cache becomes full (e.g. find and remove the least [recently] used entry) For this WL entry we will use a cache of infinite size. The reasoning behind this is that: - is is easy to do: we have temporary tables that can grow to arbitrarily large size while still providing the same insert/lookup interface. - it suits us: unless the subquery is resolved with one index lookup, hitting the cache would be many times cheaper than re-running the subquery, so cache is worth having. 4. Interplay with other subquery optimizations ---------------------------------------------- * This WL entry should not care about IN->EXISTS transformation: caching for IN subquery and result of its conversion to EXISTS would work in the same way. * This optimization is orthogonal to <=>ANY -> MIN/MAX rewrite (it will work/be useful irrespectively of whether the rewrite has been performed or not) * TODO: compare this with materialization for uncorrelated IN-subqueries. Is this basically the same? A: no, it is not: - IN-Materialization has to perform full materialization before it can do the first subquery evaluation. This WL's code has almost no startup costs. - This optimization has temp.table of (corr_reference, predicate_value), while IN-materialization will have (corr_reference) only. 5. User interface ----------------- * There will be an @@optimizer_switch flag to turn this optimization on and off (TODO: name of the flag?) * TODO: how do we show this in EXPLAIN [EXTENDED]? The most easiest is to print something in the warning text of EXPLAIN EXTEDED that would indicate use of cache. * temporary table sizing (max size for heap table, whether to use MyISAM or Maria) will be controlled with common temp.table control variables. LOW-LEVEL DESIGN: * Target version: base on mysql-5.2 code All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). How to fill the temptable ------------------------- Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() and is supposed to be quite trivial. How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. Setup process ~~~~~~~~~~~~~ Setup is performed in the same way as in setup_sj_materialization(), see the code that starts these lines: /* Create/initialize everything we will need to index lookups into the temptable. */ and ends at this line: Remove the injected semi-join IN-equalities from join_tab conds. This <questionable> We'll also need to check equalities, i.e. do an equivalent of this: if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, emb_sj_nest->sj_subq_pred))) DBUG_RETURN(TRUE); /* purecov: inspected */ Question: or perhaps that is not necessarry? </questionable> Doing the lookup ~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); The code in this WL will use the same approach Extracting the value of the subquery predicate ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The goal of making the lookup is to get the value of subquery predicate. This is done by creating an Item_field $I which refers to appropriate temporary table's field and then subquery_predicate->val_int() will invoke $I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so forth. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache (66)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache CREATION DATE..: Wed, 25 Nov 2009, 22:25 SUPERVISOR.....: Monty IMPLEMENTOR....: Sanja COPIES TO......: CATEGORY.......: Client-Sprint TASK ID........: 66 (http://askmonty.org/worklog/?tid=66) VERSION........: 9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Category updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Client-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Version updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-5.3 +9.x -=-=(Monty - Fri, 29 Jan 2010, 19:07)=-=- Version updated. --- /tmp/wklog.66.old.5893 2010-01-29 19:07:10.000000000 +0200 +++ /tmp/wklog.66.new.5893 2010-01-29 19:07:10.000000000 +0200 @@ -1 +1 @@ -Server-5.2 +Server-5.3 -=-=(Psergey - Wed, 20 Jan 2010, 14:50)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.26873 2010-01-20 14:50:41.000000000 +0200 +++ /tmp/wklog.66.new.26873 2010-01-20 14:50:41.000000000 +0200 @@ -4,7 +4,6 @@ To check/discuss: ----------------- -* Do we put subquery cache on all levels of subqueries or on highest level only * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* @@ -41,7 +40,12 @@ - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. -We dont support cases when outer_expr or correlation_references are blobs. +We don't support cases when outer_expr or correlation_references are blobs. + +All subquery predicates are cached. That is, if one subquery predicate is +located within another, both of them will have caches (one option to reduce +cache memory usage was to use cache only for the upper-most select. we decided +against it). 2. Data structure used for the cache ------------------------------------ -=-=(Psergey - Wed, 20 Jan 2010, 13:07)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.17649 2010-01-20 13:07:07.000000000 +0200 +++ /tmp/wklog.66.new.17649 2010-01-20 13:07:07.000000000 +0200 @@ -3,7 +3,13 @@ To check/discuss: - To put subquery cache on all levels of subqueries or on highest level only. +----------------- +* Do we put subquery cache on all levels of subqueries or on highest level only +* Will there be any means to measure subquery cache hit rate? +* MySQL-6.0 has a one-element predicate result cache. It is called "left + expression cache", grep for left_expr_cache in sql/item_subselect.* + When this WL is merged with 6.0's optimizations, these two caches will + need to be unified somehow. <contents> -=-=(Psergey - Mon, 18 Jan 2010, 16:40)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24899 2010-01-18 16:40:16.000000000 +0200 +++ /tmp/wklog.66.new.24899 2010-01-18 16:40:16.000000000 +0200 @@ -1,3 +1,5 @@ +* Target version: base on mysql-5.2 code + All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) -=-=(Psergey - Mon, 18 Jan 2010, 16:37)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24586 2010-01-18 16:37:07.000000000 +0200 +++ /tmp/wklog.66.new.24586 2010-01-18 16:37:07.000000000 +0200 @@ -4,6 +4,11 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to fill the temptable +------------------------- +Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() +and is supposed to be quite trivial. + How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. -=-=(Psergey - Mon, 18 Jan 2010, 16:34)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24328 2010-01-18 16:34:19.000000000 +0200 +++ /tmp/wklog.66.new.24328 2010-01-18 16:34:19.000000000 +0200 @@ -32,8 +32,8 @@ Question: or perhaps that is not necessarry? </questionable> -Execution process -~~~~~~~~~~~~~~~~~ +Doing the lookup +~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ @@ -42,4 +42,12 @@ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); +The code in this WL will use the same approach +Extracting the value of the subquery predicate +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The goal of making the lookup is to get the value of subquery predicate. +This is done by creating an Item_field $I which refers to appropriate +temporary table's field and then subquery_predicate->val_int() will invoke +$I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so +forth. -=-=(Psergey - Mon, 18 Jan 2010, 16:23)=-=- Low Level Design modified. --- /tmp/wklog.66.old.23203 2010-01-18 16:23:18.000000000 +0200 +++ /tmp/wklog.66.new.23203 2010-01-18 16:23:18.000000000 +0200 @@ -31,3 +31,15 @@ Question: or perhaps that is not necessarry? </questionable> + +Execution process +~~~~~~~~~~~~~~~~~ +SJ-Materialization does lookup in sub_select_sjm(), with this code: + + /* Do index lookup in the materialized table */ + if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) + DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ + if (res || !sjm->in_equality->val_int()) + DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); + + -=-=(Psergey - Mon, 18 Jan 2010, 16:22)=-=- Low Level Design modified. --- /tmp/wklog.66.old.23076 2010-01-18 16:22:07.000000000 +0200 +++ /tmp/wklog.66.new.23076 2010-01-18 16:22:07.000000000 +0200 @@ -4,3 +4,30 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to make lookups into temptable +---------------------------------- +We'll reuse approach used by SJ-Materialization in 6.0. + +Setup process +~~~~~~~~~~~~~ +Setup is performed in the same way as in setup_sj_materialization(), +see the code that starts these lines: + + /* + Create/initialize everything we will need to index lookups into the + temptable. + */ + +and ends at this line: + + Remove the injected semi-join IN-equalities from join_tab conds. This + +<questionable> +We'll also need to check equalities, i.e. do an equivalent of this: + + if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, + emb_sj_nest->sj_subq_pred))) + DBUG_RETURN(TRUE); /* purecov: inspected */ + +Question: or perhaps that is not necessarry? +</questionable> ------------------------------------------------------------ -=-=(View All Progress Notes, 15 total)=-=- http://askmonty.org/worklog/index.pl?tid=66&nolimit=1 DESCRIPTION: Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and return cached result of subquery if the reference values combination has already been used. For example in the following subquery: (L1, L2) IN (SELECT A, B FROM T WHERE T.F1>OTER_FIELD) set of references to look into the subquery cache is (L1, L2, OTER_FIELD). The subquery cache should be implemented as simple LRU connected to the subquery. Size of the subquery cache (in number of results (but maybe in used memory amount)) is limited by session variable (query parameter?). HIGH-LEVEL SPECIFICATION: Attach subquery cache to each Item_subquery. Interface should allow to use hash or temporary table inside. To check/discuss: ----------------- * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* When this WL is merged with 6.0's optimizations, these two caches will need to be unified somehow. <contents> 1. Scope of the task 2. Data structure used for the cache 3. Cache size 4. Interplay with other subquery optimizations 5. User interface </contents> 1. Scope of the task -------------------- This WL should handle all subquery predicates, i.e. it should handle these cases: outer_expr IN (SELECT correlated_select) outer_expr $CMP$ ALL/ANY (SELECT correlated_select) EXISTS (SELECT correlated_select) scalar-context subquery: (SELECT correlated_select) The cache will maintain (outer_expr, correlation_references)-> subquery_item_result mapping, where - correlation_references is a list of tablename.column_name that are referred from the correlated_select but tablename is a table that is ouside the subquery. - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. We don't support cases when outer_expr or correlation_references are blobs. All subquery predicates are cached. That is, if one subquery predicate is located within another, both of them will have caches (one option to reduce cache memory usage was to use cache only for the upper-most select. we decided against it). 2. Data structure used for the cache ------------------------------------ There are two data structures available in the codebase that will allow fast equality lookups: 1. HASH (mysys/hash.c) tables 2. Temporary tables (the ones that are used for e.g. GROUP BY) None of them has any support for element eviction on overflow (using LRU or some other policy). Query cache and MyISAM/Maria's key/page cache ought to support some eviction mechanism, but code-wise it is not readily reusable, one will need to factor it out (or copy it). We choose to use #2, and not to have any eviction policy. See subsequent sections for details and reasoning behind the decision. 3. Cache size ------------- Typically, a cache has some maximum size and a policy which is used to select a cache entry for removal when the cache becomes full (e.g. find and remove the least [recently] used entry) For this WL entry we will use a cache of infinite size. The reasoning behind this is that: - is is easy to do: we have temporary tables that can grow to arbitrarily large size while still providing the same insert/lookup interface. - it suits us: unless the subquery is resolved with one index lookup, hitting the cache would be many times cheaper than re-running the subquery, so cache is worth having. 4. Interplay with other subquery optimizations ---------------------------------------------- * This WL entry should not care about IN->EXISTS transformation: caching for IN subquery and result of its conversion to EXISTS would work in the same way. * This optimization is orthogonal to <=>ANY -> MIN/MAX rewrite (it will work/be useful irrespectively of whether the rewrite has been performed or not) * TODO: compare this with materialization for uncorrelated IN-subqueries. Is this basically the same? A: no, it is not: - IN-Materialization has to perform full materialization before it can do the first subquery evaluation. This WL's code has almost no startup costs. - This optimization has temp.table of (corr_reference, predicate_value), while IN-materialization will have (corr_reference) only. 5. User interface ----------------- * There will be an @@optimizer_switch flag to turn this optimization on and off (TODO: name of the flag?) * TODO: how do we show this in EXPLAIN [EXTENDED]? The most easiest is to print something in the warning text of EXPLAIN EXTEDED that would indicate use of cache. * temporary table sizing (max size for heap table, whether to use MyISAM or Maria) will be controlled with common temp.table control variables. LOW-LEVEL DESIGN: * Target version: base on mysql-5.2 code All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). How to fill the temptable ------------------------- Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() and is supposed to be quite trivial. How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. Setup process ~~~~~~~~~~~~~ Setup is performed in the same way as in setup_sj_materialization(), see the code that starts these lines: /* Create/initialize everything we will need to index lookups into the temptable. */ and ends at this line: Remove the injected semi-join IN-equalities from join_tab conds. This <questionable> We'll also need to check equalities, i.e. do an equivalent of this: if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, emb_sj_nest->sj_subq_pred))) DBUG_RETURN(TRUE); /* purecov: inspected */ Question: or perhaps that is not necessarry? </questionable> Doing the lookup ~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); The code in this WL will use the same approach Extracting the value of the subquery predicate ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The goal of making the lookup is to get the value of subquery predicate. This is done by creating an Item_field $I which refers to appropriate temporary table's field and then subquery_predicate->val_int() will invoke $I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so forth. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache (66)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Subquery optimization: Avoid recalculating subquery if external fields values found in subquery cache CREATION DATE..: Wed, 25 Nov 2009, 22:25 SUPERVISOR.....: Monty IMPLEMENTOR....: Sanja COPIES TO......: CATEGORY.......: Client-Sprint TASK ID........: 66 (http://askmonty.org/worklog/?tid=66) VERSION........: 9.x STATUS.........: Assigned PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Category updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-BackLog +Client-Sprint -=-=(Igor - Wed, 10 Mar 2010, 21:12)=-=- Version updated. --- /tmp/wklog.66.old.31558 2010-03-10 21:12:50.000000000 +0000 +++ /tmp/wklog.66.new.31558 2010-03-10 21:12:50.000000000 +0000 @@ -1 +1 @@ -Server-5.3 +9.x -=-=(Monty - Fri, 29 Jan 2010, 19:07)=-=- Version updated. --- /tmp/wklog.66.old.5893 2010-01-29 19:07:10.000000000 +0200 +++ /tmp/wklog.66.new.5893 2010-01-29 19:07:10.000000000 +0200 @@ -1 +1 @@ -Server-5.2 +Server-5.3 -=-=(Psergey - Wed, 20 Jan 2010, 14:50)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.26873 2010-01-20 14:50:41.000000000 +0200 +++ /tmp/wklog.66.new.26873 2010-01-20 14:50:41.000000000 +0200 @@ -4,7 +4,6 @@ To check/discuss: ----------------- -* Do we put subquery cache on all levels of subqueries or on highest level only * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* @@ -41,7 +40,12 @@ - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. -We dont support cases when outer_expr or correlation_references are blobs. +We don't support cases when outer_expr or correlation_references are blobs. + +All subquery predicates are cached. That is, if one subquery predicate is +located within another, both of them will have caches (one option to reduce +cache memory usage was to use cache only for the upper-most select. we decided +against it). 2. Data structure used for the cache ------------------------------------ -=-=(Psergey - Wed, 20 Jan 2010, 13:07)=-=- High-Level Specification modified. --- /tmp/wklog.66.old.17649 2010-01-20 13:07:07.000000000 +0200 +++ /tmp/wklog.66.new.17649 2010-01-20 13:07:07.000000000 +0200 @@ -3,7 +3,13 @@ To check/discuss: - To put subquery cache on all levels of subqueries or on highest level only. +----------------- +* Do we put subquery cache on all levels of subqueries or on highest level only +* Will there be any means to measure subquery cache hit rate? +* MySQL-6.0 has a one-element predicate result cache. It is called "left + expression cache", grep for left_expr_cache in sql/item_subselect.* + When this WL is merged with 6.0's optimizations, these two caches will + need to be unified somehow. <contents> -=-=(Psergey - Mon, 18 Jan 2010, 16:40)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24899 2010-01-18 16:40:16.000000000 +0200 +++ /tmp/wklog.66.new.24899 2010-01-18 16:40:16.000000000 +0200 @@ -1,3 +1,5 @@ +* Target version: base on mysql-5.2 code + All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) -=-=(Psergey - Mon, 18 Jan 2010, 16:37)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24586 2010-01-18 16:37:07.000000000 +0200 +++ /tmp/wklog.66.new.24586 2010-01-18 16:37:07.000000000 +0200 @@ -4,6 +4,11 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to fill the temptable +------------------------- +Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() +and is supposed to be quite trivial. + How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. -=-=(Psergey - Mon, 18 Jan 2010, 16:34)=-=- Low Level Design modified. --- /tmp/wklog.66.old.24328 2010-01-18 16:34:19.000000000 +0200 +++ /tmp/wklog.66.new.24328 2010-01-18 16:34:19.000000000 +0200 @@ -32,8 +32,8 @@ Question: or perhaps that is not necessarry? </questionable> -Execution process -~~~~~~~~~~~~~~~~~ +Doing the lookup +~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ @@ -42,4 +42,12 @@ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); +The code in this WL will use the same approach +Extracting the value of the subquery predicate +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The goal of making the lookup is to get the value of subquery predicate. +This is done by creating an Item_field $I which refers to appropriate +temporary table's field and then subquery_predicate->val_int() will invoke +$I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so +forth. -=-=(Psergey - Mon, 18 Jan 2010, 16:23)=-=- Low Level Design modified. --- /tmp/wklog.66.old.23203 2010-01-18 16:23:18.000000000 +0200 +++ /tmp/wklog.66.new.23203 2010-01-18 16:23:18.000000000 +0200 @@ -31,3 +31,15 @@ Question: or perhaps that is not necessarry? </questionable> + +Execution process +~~~~~~~~~~~~~~~~~ +SJ-Materialization does lookup in sub_select_sjm(), with this code: + + /* Do index lookup in the materialized table */ + if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) + DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ + if (res || !sjm->in_equality->val_int()) + DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); + + -=-=(Psergey - Mon, 18 Jan 2010, 16:22)=-=- Low Level Design modified. --- /tmp/wklog.66.old.23076 2010-01-18 16:22:07.000000000 +0200 +++ /tmp/wklog.66.new.23076 2010-01-18 16:22:07.000000000 +0200 @@ -4,3 +4,30 @@ Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). +How to make lookups into temptable +---------------------------------- +We'll reuse approach used by SJ-Materialization in 6.0. + +Setup process +~~~~~~~~~~~~~ +Setup is performed in the same way as in setup_sj_materialization(), +see the code that starts these lines: + + /* + Create/initialize everything we will need to index lookups into the + temptable. + */ + +and ends at this line: + + Remove the injected semi-join IN-equalities from join_tab conds. This + +<questionable> +We'll also need to check equalities, i.e. do an equivalent of this: + + if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, + emb_sj_nest->sj_subq_pred))) + DBUG_RETURN(TRUE); /* purecov: inspected */ + +Question: or perhaps that is not necessarry? +</questionable> ------------------------------------------------------------ -=-=(View All Progress Notes, 15 total)=-=- http://askmonty.org/worklog/index.pl?tid=66&nolimit=1 DESCRIPTION: Collect all outer items/references (left part of the subquiery and outer references inside the subquery) in key string. Compare the string (which represents certain value set of the references) against values in hash table and return cached result of subquery if the reference values combination has already been used. For example in the following subquery: (L1, L2) IN (SELECT A, B FROM T WHERE T.F1>OTER_FIELD) set of references to look into the subquery cache is (L1, L2, OTER_FIELD). The subquery cache should be implemented as simple LRU connected to the subquery. Size of the subquery cache (in number of results (but maybe in used memory amount)) is limited by session variable (query parameter?). HIGH-LEVEL SPECIFICATION: Attach subquery cache to each Item_subquery. Interface should allow to use hash or temporary table inside. To check/discuss: ----------------- * Will there be any means to measure subquery cache hit rate? * MySQL-6.0 has a one-element predicate result cache. It is called "left expression cache", grep for left_expr_cache in sql/item_subselect.* When this WL is merged with 6.0's optimizations, these two caches will need to be unified somehow. <contents> 1. Scope of the task 2. Data structure used for the cache 3. Cache size 4. Interplay with other subquery optimizations 5. User interface </contents> 1. Scope of the task -------------------- This WL should handle all subquery predicates, i.e. it should handle these cases: outer_expr IN (SELECT correlated_select) outer_expr $CMP$ ALL/ANY (SELECT correlated_select) EXISTS (SELECT correlated_select) scalar-context subquery: (SELECT correlated_select) The cache will maintain (outer_expr, correlation_references)-> subquery_item_result mapping, where - correlation_references is a list of tablename.column_name that are referred from the correlated_select but tablename is a table that is ouside the subquery. - subquery_item_result is 'bool' for subquery predicates, and is of some scalar or ROW(scalar1,...scalarN) type for scalar-context subquery. We don't support cases when outer_expr or correlation_references are blobs. All subquery predicates are cached. That is, if one subquery predicate is located within another, both of them will have caches (one option to reduce cache memory usage was to use cache only for the upper-most select. we decided against it). 2. Data structure used for the cache ------------------------------------ There are two data structures available in the codebase that will allow fast equality lookups: 1. HASH (mysys/hash.c) tables 2. Temporary tables (the ones that are used for e.g. GROUP BY) None of them has any support for element eviction on overflow (using LRU or some other policy). Query cache and MyISAM/Maria's key/page cache ought to support some eviction mechanism, but code-wise it is not readily reusable, one will need to factor it out (or copy it). We choose to use #2, and not to have any eviction policy. See subsequent sections for details and reasoning behind the decision. 3. Cache size ------------- Typically, a cache has some maximum size and a policy which is used to select a cache entry for removal when the cache becomes full (e.g. find and remove the least [recently] used entry) For this WL entry we will use a cache of infinite size. The reasoning behind this is that: - is is easy to do: we have temporary tables that can grow to arbitrarily large size while still providing the same insert/lookup interface. - it suits us: unless the subquery is resolved with one index lookup, hitting the cache would be many times cheaper than re-running the subquery, so cache is worth having. 4. Interplay with other subquery optimizations ---------------------------------------------- * This WL entry should not care about IN->EXISTS transformation: caching for IN subquery and result of its conversion to EXISTS would work in the same way. * This optimization is orthogonal to <=>ANY -> MIN/MAX rewrite (it will work/be useful irrespectively of whether the rewrite has been performed or not) * TODO: compare this with materialization for uncorrelated IN-subqueries. Is this basically the same? A: no, it is not: - IN-Materialization has to perform full materialization before it can do the first subquery evaluation. This WL's code has almost no startup costs. - This optimization has temp.table of (corr_reference, predicate_value), while IN-materialization will have (corr_reference) only. 5. User interface ----------------- * There will be an @@optimizer_switch flag to turn this optimization on and off (TODO: name of the flag?) * TODO: how do we show this in EXPLAIN [EXTENDED]? The most easiest is to print something in the warning text of EXPLAIN EXTEDED that would indicate use of cache. * temporary table sizing (max size for heap table, whether to use MyISAM or Maria) will be controlled with common temp.table control variables. LOW-LEVEL DESIGN: * Target version: base on mysql-5.2 code All items on which subquery depend could be collected in st_select_lex::mark_as_dependent (direct of indirect reference?) Temporary table index should be created by all fields except result field (TMP_TABLE_PARAM::keyinfo). How to fill the temptable ------------------------- Can reuse approach from SJ-Materialization. Its code is in end_sj_materialize() and is supposed to be quite trivial. How to make lookups into temptable ---------------------------------- We'll reuse approach used by SJ-Materialization in 6.0. Setup process ~~~~~~~~~~~~~ Setup is performed in the same way as in setup_sj_materialization(), see the code that starts these lines: /* Create/initialize everything we will need to index lookups into the temptable. */ and ends at this line: Remove the injected semi-join IN-equalities from join_tab conds. This <questionable> We'll also need to check equalities, i.e. do an equivalent of this: if (!(sjm->in_equality= create_subq_in_equalities(thd, sjm, emb_sj_nest->sj_subq_pred))) DBUG_RETURN(TRUE); /* purecov: inspected */ Question: or perhaps that is not necessarry? </questionable> Doing the lookup ~~~~~~~~~~~~~~~~ SJ-Materialization does lookup in sub_select_sjm(), with this code: /* Do index lookup in the materialized table */ if ((res= join_read_key2(join_tab, sjm->table, sjm->tab_ref)) == 1) DBUG_RETURN(NESTED_LOOP_ERROR); /* purecov: inspected */ if (res || !sjm->in_equality->val_int()) DBUG_RETURN(NESTED_LOOP_NO_MORE_ROWS); The code in this WL will use the same approach Extracting the value of the subquery predicate ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The goal of making the lookup is to get the value of subquery predicate. This is done by creating an Item_field $I which refers to appropriate temporary table's field and then subquery_predicate->val_int() will invoke $I->val_int(), subquery_predicate->val_str() will invoke $I->val_str() and so forth. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport subquery optimizations (104)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport subquery optimizations CREATION DATE..: Wed, 10 Mar 2010, 18:54 SUPERVISOR.....: Igor IMPLEMENTOR....: Psergey COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 104 (http://askmonty.org/worklog/?tid=104) VERSION........: Server-5.3 STATUS.........: Complete PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 20:55)=-=- High Level Description modified. --- /tmp/wklog.104.old.30849 2010-03-10 20:55:30.000000000 +0000 +++ /tmp/wklog.104.new.30849 2010-03-10 20:55:30.000000000 +0000 @@ -1,2 +1,2 @@ -The target of this task is backport the code for subquery optimizations from the +The goal of this task is to backport the code for subquery optimizations from the MySQL 6.0 code line to MariaDB 5.3. -=-=(Igor - Wed, 10 Mar 2010, 20:54)=-=- High Level Description modified. --- /tmp/wklog.104.old.30758 2010-03-10 20:54:41.000000000 +0000 +++ /tmp/wklog.104.new.30758 2010-03-10 20:54:41.000000000 +0000 @@ -1,2 +1,2 @@ The target of this task is backport the code for subquery optimizations from the -MySQL 6.0 code line to MariaDB 5.3 +MySQL 6.0 code line to MariaDB 5.3. -=-=(Igor - Wed, 10 Mar 2010, 20:53)=-=- Title modified. --- /tmp/wklog.104.old.30184 2010-03-10 20:53:31.000000000 +0000 +++ /tmp/wklog.104.new.30184 2010-03-10 20:53:31.000000000 +0000 @@ -1 +1 @@ -Backport 6.0 subquery code +Backport subquery optimizations -=-=(Igor - Wed, 10 Mar 2010, 20:52)=-=- High Level Description modified. --- /tmp/wklog.104.old.30172 2010-03-10 20:52:27.000000000 +0000 +++ /tmp/wklog.104.new.30172 2010-03-10 20:52:27.000000000 +0000 @@ -1 +1,2 @@ -Backport 6.0 subquery code to MariaDB 5.3 +The target of this task is backport the code for subquery optimizations from the +MySQL 6.0 code line to MariaDB 5.3 -=-=(Guest - Wed, 10 Mar 2010, 20:48)=-=- Title modified. --- /tmp/wklog.104.old.29904 2010-03-10 20:48:13.000000000 +0000 +++ /tmp/wklog.104.new.29904 2010-03-10 20:48:13.000000000 +0000 @@ -1 +1 @@ -Backport 6.0 subquery code to MariaDB 5.3 +Backport 6.0 subquery code -=-=(Psergey - Wed, 10 Mar 2010, 18:55)=-=- Dependency created: 91 now depends on 104 DESCRIPTION: The goal of this task is to backport the code for subquery optimizations from the MySQL 6.0 code line to MariaDB 5.3. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Igor): Backport subquery optimizations (104)
by worklog-noreply＠askmonty.org 10 Mar '10

10 Mar '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Backport subquery optimizations CREATION DATE..: Wed, 10 Mar 2010, 18:54 SUPERVISOR.....: Igor IMPLEMENTOR....: Psergey COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 104 (http://askmonty.org/worklog/?tid=104) VERSION........: Server-5.3 STATUS.........: Complete PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Igor - Wed, 10 Mar 2010, 20:55)=-=- High Level Description modified. --- /tmp/wklog.104.old.30849 2010-03-10 20:55:30.000000000 +0000 +++ /tmp/wklog.104.new.30849 2010-03-10 20:55:30.000000000 +0000 @@ -1,2 +1,2 @@ -The target of this task is backport the code for subquery optimizations from the +The goal of this task is to backport the code for subquery optimizations from the MySQL 6.0 code line to MariaDB 5.3. -=-=(Igor - Wed, 10 Mar 2010, 20:54)=-=- High Level Description modified. --- /tmp/wklog.104.old.30758 2010-03-10 20:54:41.000000000 +0000 +++ /tmp/wklog.104.new.30758 2010-03-10 20:54:41.000000000 +0000 @@ -1,2 +1,2 @@ The target of this task is backport the code for subquery optimizations from the -MySQL 6.0 code line to MariaDB 5.3 +MySQL 6.0 code line to MariaDB 5.3. -=-=(Igor - Wed, 10 Mar 2010, 20:53)=-=- Title modified. --- /tmp/wklog.104.old.30184 2010-03-10 20:53:31.000000000 +0000 +++ /tmp/wklog.104.new.30184 2010-03-10 20:53:31.000000000 +0000 @@ -1 +1 @@ -Backport 6.0 subquery code +Backport subquery optimizations -=-=(Igor - Wed, 10 Mar 2010, 20:52)=-=- High Level Description modified. --- /tmp/wklog.104.old.30172 2010-03-10 20:52:27.000000000 +0000 +++ /tmp/wklog.104.new.30172 2010-03-10 20:52:27.000000000 +0000 @@ -1 +1,2 @@ -Backport 6.0 subquery code to MariaDB 5.3 +The target of this task is backport the code for subquery optimizations from the +MySQL 6.0 code line to MariaDB 5.3 -=-=(Guest - Wed, 10 Mar 2010, 20:48)=-=- Title modified. --- /tmp/wklog.104.old.29904 2010-03-10 20:48:13.000000000 +0000 +++ /tmp/wklog.104.new.29904 2010-03-10 20:48:13.000000000 +0000 @@ -1 +1 @@ -Backport 6.0 subquery code to MariaDB 5.3 +Backport 6.0 subquery code -=-=(Psergey - Wed, 10 Mar 2010, 18:55)=-=- Dependency created: 91 now depends on 104 DESCRIPTION: The goal of this task is to backport the code for subquery optimizations from the MySQL 6.0 code line to MariaDB 5.3. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0