- developers - lists.mariadb.org

[Maria-developers] Getting restarted with mariadb and a port to AIX
by Michael Felt 28 Mar '16

28 Mar '16

Hi all, This week I added a few issues that I felt were clear enough to report directly. This one I am not so sure of as it concerns two areas I know too little about: a) cmake (how to get an equivalent of GNU autotools ./configure --help) b) cmake (how to make it show the commands it is actually running (so I could add -bnoquiet and/or -bloadmap:XXX.map) c) 'strategic' grepping through C++ sources to find where an undefined label is calling something. After I 'patch' aka bandaid mariadb-5.5.48.0/sql-common/mysql_async.c (see https://jira.mariadb.org/browse/MDEV-9814) I get loader errors - and I cannot, in ways I could with an autotools generated 'make' see how to debug. Q) are there variables (e.g., V=1) I can pass to make so that the commands it performs are more verbose Q) have any ideas why these symbols are undefined? Thanks, Michael

2 1

[Maria-developers] Unique Index for blob
by Sachin Setia 28 Mar '16

28 Mar '16

Hello Devlopers Hi this is sachin. I applied for gsoc 2016(applied by name sachin setiya) My topic was unique index for blob(MyISAM) But i already completed some part of it .it is working for blobs varchar for definitions like create table tbl(abc varchar(2000) , xyz blob ,unique(abc,xyz)); I have corrected the table2myisam function . For documentation you can refer to my gsoc proposal If am doing wrong please let me know. I am attaching the patch file TODO Second thing which i need to do is to tweak check_definition function to correctly check for definition currently I am commenting the function code Regards sachin

1 1

Re: [Maria-developers] Gsoc 2016 Mdev 371 Unique index for blob
by Sachin Setia 28 Mar '16

28 Mar '16

I am Sorry I did not see this mail.So I did not send the prototype link in proposal On Thu, Mar 24, 2016 at 10:41 PM, Sergei Golubchik <serg(a)mariadb.org> wrote: > Hi, Sachin! > > As you might have noticed on the mailing list, there is another student > who wants to do this project. > > It would help us choose, if you'd put your prototype code on github and > put the link to the repository into your final proposal. > > Thanks! > > On Mar 15, Sachin Setia wrote: > > Dear Developers, > > > > I was doing prototype for this project as Mr Sergei Golubchik suggested > > > > Regards, > Sergei > Chief Architect MariaDB > and security(a)mariadb.org >

1 0

[Maria-developers] Please review MDEV-6353 my_ismbchar() and my_mbcharlen() refactoring
by Alexander Barkov 28 Mar '16

28 Mar '16

Hi Sergei, Please review a partial patch for MDEV-6353. It removes my_mbcharlen() in all remaining pieces of the code, except LOAD DATA and SELECT INTO OUTFILE. I'll do LOAD DATA and SELECT INTO OUTFILE in a separate patch. Thanks.

1 0

Re: [Maria-developers] Google Summer of Code 2016
by Soham Mankad 26 Mar '16

26 Mar '16

Respected sir, I am Soham Mankad and thanks Sergei sir for directing to me this developers list. As I had mailed you before I have gone through the basic requirements and goals of the project "Aggregate Stored Functions" and also thinking on varoius syntax and languages that can be used like hsql, Postgresql, or mysql. Also as you had mentioned in a reply to Vikas Gupta still i have not found any data base with this facility existing uptill now. Also I have seen the syntax and possible ways given by you in knowledge base of the project. The most basic form that you have mentioned i.e using nested commands i.e returning sum adding two numbers. I am also learning about various aspects of create function commands in different languages loke read only or returning which rows or coloumns so that optimum result will be given. I have submitted my proposal for this project on Gsoc'16 website. I would definitely appreciate an opportunity to work on this project. Regards, Soham Mankad.

1 0

[Maria-developers] Window functions: why some functions are not in select_lex->window_funcs?
by Sergey Petrunia 24 Mar '16

24 Mar '16

Hello Igor, I was trying to move JOIN::process_window_functions() from using join->fields_list to using join->select_lex->window_funcs. And I discovered that join->select_lex->window_funcs does not contain window function items that use window definition. Is this intentional? Example: create table t0 (a int); insert into t0 values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9); create table t1 (pk int, c int); insert into t1 select a+1,1 from t0; update t1 set c=2 where pk not in (1,2,3,4); select pk, c, count(*) over (partition by c order by pk rows between 2 preceding and 2 following) as CNT from t1 Here: #1 0x0000555555af5db6 in AGGR_OP::end_send (this=0x7fff4c0086c8) (gdb) p join->select_lex->window_funcs $69 = {<base_list> = {<Sql_alloc> = {<No data fields>}, first = 0x7fff4c005d18, last = 0x7fff4c005d18, elements = 1}, <No data fields>} But: select pk, c, count(*) over w1 as CNT from t1 window w1 as (partition by c order by pk rows between 2 preceding and 2 following); (gdb) p join->select_lex->window_funcs $74 = {<base_list> = {<Sql_alloc> = {<No data fields>}, first = 0x555556d48820, last = 0x55555ab10b70, elements = 0}, <No data fields>} BR Sergei -- Sergei Petrunia, Software Developer MariaDB Corporation | Skype: sergefp | Blog: http://s.petrunia.net/blog

1 1

Re: [Maria-developers] Gsoc 2016 Mdev 371 Unique index for blob
by Sachin Setia 23 Mar '16

23 Mar '16

Hello Sergei Actually I was prototyping for blob and varchar for aria and myisam storage engine. My prototype worked for complex definations like craete table(abc int primary key, blob_col blob unique, varchar_col varchar(1000) unique) engine=myisam; Solved the issue of frm file incosistance. As you suggested for doing it for innodb i am current working on it.Innodb does not natively support hash based index. when we run select distinct column from tbl; it use create_internal_tmp_table() which uses maria storage engine for creating tmp table. But query like this works MariaDB [sachin]> create table iu2(abc blob unique); Query OK, 0 rows affected (0.04 sec) MariaDB [sachin]> insert into iu2 values(1); Query OK, 1 row affected (0.03 sec) MariaDB [sachin]> insert into iu2 values(1); ERROR 1062 (23000): Duplicate entry '1' for key 'abc' this query does not use hash but it simply compares values Will write a proposal shortly. On Tue, Mar 22, 2016 at 4:20 PM, Sachin Setia <sachinsetia1001(a)gmail.com> wrote: > Hello Sergi > Actually I was prototyping for blob and varchar for aria and myisam > storage engine. > My prototype worked for complex definations like > craete table(abc int primary key, blob_col blob unique, varchar_col > varchar(1000) unique) engine=myisam; > Solved the issue of frm file incosistance. > > As you suggested for doing it for innodb i am current working on it.Innodb > does not natively support hash based index. > when we run select distinct column from tbl; > it use create_internal_tmp_table() which uses maria storage engine for > creating tmp table. > But query like this works > MariaDB [sachin]> create table iu2(abc blob unique); > Query OK, 0 rows affected (0.04 sec) > > MariaDB [sachin]> insert into iu2 values(1); > Query OK, 1 row affected (0.03 sec) > > MariaDB [sachin]> insert into iu2 values(1); > ERROR 1062 (23000): Duplicate entry '1' for key 'abc' > this query does not use hash but it simply compares values > Will write a proposal shortly. > > > Regards > sachin > > On Sat, Mar 19, 2016 at 1:52 AM, Sergei Golubchik <serg(a)mariadb.org> > wrote: > >> Hi, Sachin! >> >> On Mar 18, Sachin Setia wrote: >> > >> > ERROR 1030 (HY000): Got error 190 "Incompatible key or row definition >> > between the MariaDB .frm file and the information in the storage engine. >> > You have to dump an" from storage engine MyISAM >> > >> > We are getting this becuase in mi_create for each unique_def it creates >> one >> > keydef and writes it.And this creates two problem >> > 1. frm keydef algorithm is different from saved kefdef algorithm(always >> > zero) for the time I have solved this problem . >> > >> > 2.Mismatch between keydef's keysegs the reason for this is when >> mi_create >> > creates keyseg for unique_def it did not keeps the orignal uniquedef's >> > keyseg parameters in mind like language start length which creates >> problem >> > in check_definition function in ha_myisam.cc.I am currently working on >> it >> > Once again sorry for this foolish mistake. >> > Regars >> > sachin >> >> No problem, everyone makes mistakes :) >> >> It's a prototype, after all. It's much more important that you >> understand how the code works and why it doesn't work. >> >> Regards, >> Sergei >> Chief Architect MariaDB >> and security(a)mariadb.org >> > >

2 3

[Maria-developers] More memory allocation for key seg
by Sachin Setia 23 Mar '16

23 Mar '16

Hello Devlopers I was debugging the source code of mariadb. My query was MariaDB [sachin]> create table pro2(abc int primary key,xyz int , ass int ,unique(xyz,ass))engine=myisam; In table2myisam function of file ha_myisam.cc i found that we are allocating more memory for keysegments if (!(my_multi_malloc(MYF(MY_WME), recinfo_out, (share->fields * 2 + 2) * sizeof(MI_COLUMNDEF), keydef_out, share->keys * sizeof(MI_KEYDEF), &keyseg, (share->key_parts + share->keys) * sizeof(HA_KEYSEG), ^ here why this NullS))) And we donot use these extra key segments because suceeding for loop use only total number of share->key_parts keysegments. And same in ha_maria file.Is this a bug or I am missing something Regards sachin

2 2

[Maria-developers] Please review MDEV-9604 crash in Item::save_in_field with empty enum value
by Alexander Barkov 22 Mar '16

22 Mar '16

Hi Sergei, It fixes the reported crash, and additionally inconsistencies when comparing temporal columns to empty strings: In some cases it returned rows, in other cases it returned no rows, depending on: - the engine - the type of comparison arguments (columns vs literals). - presence of an index Now everything works in the same way. Note, this additional problem is not 10.1 specific, it existed in 10.0. But 10.0 and 10.1 worked differently in certain cases. Thanks!

2 4

Re: [Maria-developers] Aggregate stored functions [MDEV-7773]
by Sergei Golubchik 22 Mar '16

22 Mar '16

Hi, Varun! On Mar 21, Varun Gupta wrote: > Hi Sergei, > > 4) :create an Item_sum_sp to be able to use aggregate stored functions in > queries. > Does this mean we can have the aggregate stored functions in the SELECT , > WHERE etc clauses?can you give me more details about this. Yes, of course. That's how functions are used, right? :) SELECT AVG(x), COUNT(y), MAX(z) FROM t1 WHERE aaa=12 GROUP BY bbb; for example. Regards, Sergei Chief Architect MariaDB and security(a)mariadb.org

2 3