- developers - lists.mariadb.org

[Maria-developers] MariaDB 5.1.60 and 5.2.10
by Daniel Bartholomew 06 Dec '11

06 Dec '11

All, MariaDB 5.1.60 is nearly ready to release. There are only a couple of mirrors which don't have the files. The changelog and release notes pages are here: * http://kb.askmonty.org/en/mariadb-5160-release-notes * http://kb.askmonty.org/en/mariadb-5160-changelog Unless something comes up, I will probably activate the download page for this release and 5.2.10 in a couple of hours. This will be followed by a combined announce email and so forth. Thanks. -- Daniel Bartholomew MariaDB - http://mariadb.org Monty Program - http://montyprogram.com AskMonty Knowledgebase - http://kb.askmonty.org

1 0

[Maria-developers] MariaDB 5.2.10
by Daniel Bartholomew 05 Dec '11

05 Dec '11

All, The MariaDB 5.2.10 release is just about ready to go. Most of the packages have been mirrored. We're still waiting on Windows packages (we're having some issues with the Windows build server), but they will hopefully be available soon. This brings up the question: Should we delay the release until we have Windows packages or add them after the fact? The Release Notes and Changelog for the release are here: - http://kb.askmonty.org/en/mariadb-5210-release-notes - http://kb.askmonty.org/en/mariadb-5210-changelog Please let me know ASAP if there is anything that should be mentioned in the Release Notes that currently is not (or jump in and add it directly to the page yourself). Serg: The PAM note in the release notes is pretty sparse. Can you add some more to it? Thanks. -- Daniel Bartholomew MariaDB - http://mariadb.org Monty Program - http://montyprogram.com AskMonty Knowledgebase - http://kb.askmonty.org

1 0

Re: [Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 3120: Added suppressions
by Sergei 03 Dec '11

03 Dec '11

Hi, Monty! Can you please explain how it is possible to have "different number rows in the log" ? On Dec 03, noreply(a)launchpad.net wrote: > ------------------------------------------------------------ > revno: 3120 > committer: Michael Widenius <monty(a)askmonty.org> > branch nick: maria-5.1 > timestamp: Sat 2011-12-03 20:29:15 +0200 > message: > Added suppressions > Fixed feedback_plugin_send to not generate a random number of lines. > modified: > mysql-test/t/feedback_plugin_send.test > mysql-test/valgrind.supp > support-files/compiler_warnings.supp > > === modified file 'mysql-test/t/feedback_plugin_send.test' > --- mysql-test/t/feedback_plugin_send.test 2011-11-13 12:28:35 +0000 > +++ mysql-test/t/feedback_plugin_send.test 2011-12-03 18:29:15 +0000 > @@ -17,7 +17,16 @@ > perl; > $log_error= $ENV{'MYSQLTEST_VARDIR'} . '/log/mysqld.1.err'; > open(LOG, '<', $log_error) or die "open(< $log_error): $!"; > - /feedback plugin:.*/ && print "$&\n" while $_=<LOG>; > + > + # Get the first few rows (as there may be different number rows in the log) > + $i= 0; > + while ($_=<LOG>) > + { > + if (/feedback plugin:.*/) > + { > + print "$&\n"; > + break if ($i++ >= 3); > + } > + } > close LOG; > EOF > - > Regards, Sergei

1 0

[Maria-developers] 5.3: optimizer status update and test requests
by Sergei Petrunia 01 Dec '11

01 Dec '11

Hi Elena, A quick update on what has been happening in 5.3 optimizer recently. The text is FYI, requests for QA are marked with 'REQUEST': == Semi-joins: no non-PS crashes, optimizations refactoring == When I was fixing BUG#887468, I have made some refactorings to semi-join optimization code and pushed them into 5.3-main (revno: 3316, revision-id: psergey(a)askmonty.org-20111126082752-u8s9tv928lc67c8o) This change has grown to be more than just refactoring: - EXPLAIN now shows "Start temporary" at different position (documentation pending) - The optimizer may pick certain plans that were not picked before. REQUEST#1: please re-run semi-join optimization tests for the new server. == Discovery of effects of @@optimizer_prune_level == I have (yesterday) found https://bugs.launchpad.net/maria/+bug/898747. In a nutshell, default value of @@optimizer_prune_level=1 causes the optimizer to prune away some semi-join query plans. There can be different views about how serious BUG#898747 is, but its effect is that certain range of query plans were not tested before. REQUEST#2: re-run semi-join optimization tests with @@optimizer_prune_level=0 This can be joined with REQUEST#1 (however, if you find bugs, we're interested to know whether they have been present before the revno:3316 patch). To be continued... BR Sergei -- Sergei Petrunia, Software Developer Monty Program AB, http://askmonty.org Blog: http://s.petrunia.net/blog

2 2

[Maria-developers] Proposal: change SUBQUERY to MATERIALIZE
by Sergei Petrunia 01 Dec '11

01 Dec '11

2 1

Re: [Maria-developers] Proposal: change SUBQUERY to MATERIALIZE
by Sergei Petrunia 01 Dec '11

01 Dec '11

1 0

Re: [Maria-developers] Proposal: change SUBQUERY to MATERIALIZE
by Sergei Petrunia 01 Dec '11

01 Dec '11

1 0

[Maria-developers] MYSQL_THDVAR and SET GLOBAL
by MARK CALLAGHAN 30 Nov '11

30 Nov '11

Is there an easy way to prevent SET GLOBAL ... from being done for plugin variables declared via MYSQL_THDVAR_...? For the following variable I can do SET GLOBAL innodb_fake_changes=1 and that sets a global variable. static MYSQL_THDVAR_BOOL(fake_changes, PLUGIN_VAR_OPCMDARG, "In the transaction after enabled, UPDATE, INSERT and DELETE only move the cursor to the records " "and do nothing other operations (no changes, no ibuf, no undo, no transaction log) in the transaction. " "This is to cause replication prefetch IO. ATTENTION: the transaction started after enabled is affected.", check_fake_changes, NULL, FALSE); -- Mark Callaghan mdcallag(a)gmail.com

3 2

[Maria-developers] MariaDB 5.5 binary tarball packages and libaio
by Kristian Nielsen 30 Nov '11

30 Nov '11

Hi Wlad, I got some progress with Buildbot and producing binary tarball packages for MariaDB 5.5, using your suggestions. But I encountered a problem. I used -DBUILD_CONFIG=mysql-release and installed libaio-dev. But then the server fails like this: ./bin/mysqld: error while loading shared libraries: libaio.so.1: cannot open shared object file: No such file or directory Of course - as libaio.so must be installed by users. I do not think it is appropriate that users need to install libaio.so to use the binary tarballs (as those are supposed to work anywhere - on the other hand, for .deb for example we _should_ use libaio of course, and dependencies will take care that it is installed). So what to do? For now I've tried to build without libaio for the generic tarballs (-DIGNORE_AIO_CHECK=1). So binaries will work; to get aio, users can build themselves or install real packages. Or do you have another idea? Do you know what MySQL does? One option could be statically linking libaio. Now in general, static linking is a can of worms with many pitfalls, however for just libaio it might work (libaio.so is statically linked on my Debian Wheezy). However, I am not sure if this is safe in general, nor do I know how to make cmake link just libaio statically? Any ideas? - Kristian.

4 4

Re: [Maria-developers] [Commits] Rev 3322: Fixed bug lp:825051 in file:///home/tsk/mprog/src/5.3/
by Oleksandr Byelkin 30 Nov '11

30 Nov '11

29.11.2011 23:06, timour(a)askmonty.org пишет: > At file:///home/tsk/mprog/src/5.3/ > > ------------------------------------------------------------ > revno: 3322 > revision-id: timour(a)askmonty.org-20111129210639-bjcn2pfocub6uwba > parent: holyfoot(a)askmonty.org-20111129112752-6cy8wy7h5q58l6je > fixes bug(s): https://launchpad.net/bugs/825051 > committer: timour(a)askmonty.org > branch nick: 5.3 > timestamp: Tue 2011-11-29 23:06:39 +0200 > message: > Fixed bug lp:825051 > > The cause of the wrong result was that Item_ref_null_helper::get_date() > didn't use a method of the *_result() family, and fetched the data > for the field from the current row instead of result_field. Changed to > use the correct *_result() method, like to all other similar methods > of Item_ref_null_helper. OK to push. Do not forget that we agreed to fix this typo in 5.2 at least.

1 0