February 2023 - developers - lists.mariadb.org

Re: [Maria-developers] 6e28a576913: move alloca() definition from all *.h files to one place
by Sergei Golubchik 10 Feb '23

10 Feb '23

Hi, Julius, this is the combined diff of both commits: git diff e1ce867e5ab0c^^ e1ce867e5ab0c On Feb 10, Julius Goryavsky wrote: > diff --git a/include/CMakeLists.txt b/include/CMakeLists.txt > index a7b98a11050..7d87c4a42b3 100644 > --- a/include/CMakeLists.txt > +++ b/include/CMakeLists.txt > @@ -104,6 +105,7 @@ ENDMACRO() > > INSTALL_COMPAT_HEADER(my_global.h "") > INSTALL_COMPAT_HEADER(my_config.h "") > +INSTALL_COMPAT_HEADER(my_alloca.h "") I wouldn't use INSTALL_COMPAT_HEADER for my_alloca.h > INSTALL_COMPAT_HEADER(my_sys.h "") > INSTALL_COMPAT_HEADER(mysql_version.h " > #include <mariadb_version.h> > diff --git a/include/my_alloca.h b/include/my_alloca.h > new file mode 100644 > index 00000000000..fec0320195e > --- /dev/null > +++ b/include/my_alloca.h > @@ -0,0 +1,37 @@ > +/* Copyright (c) 2023, MariaDB Corporation. > + > + This program is free software; you can redistribute it and/or modify > + it under the terms of the GNU General Public License as published by > + the Free Software Foundation; version 2 of the License. > + > + This program is distributed in the hope that it will be useful, > + but WITHOUT ANY WARRANTY; without even the implied warranty of > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > + GNU General Public License for more details. > + > + You should have received a copy of the GNU General Public License > + along with this program; if not, write to the Free Software > + Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1335 USA */ > + > +#ifndef MY_ALLOCA_INCLUDED > +#define MY_ALLOCA_INCLUDED > + > +#ifdef _WIN32 > +#include <malloc.h> /*for alloca*/ > +#ifndef alloca When can alloca be defined? > +#define alloca _alloca > +#endif > +#else > +#include <stdlib.h> why stdlib.h? > +#ifdef HAVE_ALLOCA_H > +#include <alloca.h> > +#endif > +#endif > + > +#if defined(HAVE_ALLOCA) > +#if defined(__GNUC__) && !defined(HAVE_ALLOCA_H) && !defined(alloca) add a comment describing your MinGW use case when this #if is needed > +#define alloca __builtin_alloca > +#endif /* GNUC */ > +#endif > + > +#endif /* MY_ALLOCA_INCLUDED */ > diff --git a/include/mysql/service_encryption.h b/include/mysql/service_encryption.h > index 69d205a27e8..cab1418e1d6 100644 > --- a/include/mysql/service_encryption.h > +++ b/include/mysql/service_encryption.h > @@ -24,21 +24,19 @@ > *provider* (encryption plugin). > */ > > +#ifndef MYSQL_ABI_CHECK > +#include <my_alloca.h> > +#endif > + > #ifdef __cplusplus > extern "C" { > #endif > > #ifndef MYSQL_ABI_CHECK why not to put #include <my_alloca.h> here? > #ifdef _WIN32 > -#include <malloc.h> > #ifndef __cplusplus > #define inline __inline > #endif > -#else > -#include <stdlib.h> I suspect stdlib.h belongs here, not in my_alloca.h > -#ifdef HAVE_ALLOCA_H > -#include <alloca.h> > -#endif > #endif > #endif Regards, Sergei VP of MariaDB Server Engineering and security(a)mariadb.org

1 0

Re: [Maria-developers] ceb75e9bd9e: MDEV-30203: Move mysql symlinks to different package (fixes)
by Sergei Golubchik 10 Feb '23

10 Feb '23

Hi, Daniel, On Feb 09, Daniel Black wrote: > revision-id: ceb75e9bd9e (mariadb-10.11.1-152-gceb75e9bd9e) > parent(s): 526fce28b90 > author: Daniel Black > committer: Daniel Black > timestamp: 2023-02-09 12:20:57 +1100 > message: > > MDEV-30203: Move mysql symlinks to different package (fixes) > > Fixes MDEV-30571 MariaDB-client does not pull MariaDB-compat anymore, conflicts with RHEL packages > > Fixes MDEV-30574 MariaDB-server conflicts with MariaDB-common from previous versions > > MariaDB-Server now depends on MariaDB-common 11.0+. > > MariaDB-{Client,Server} append to the Requirements rather than overriding them. > > MariaDB-Server package now depends on the MariaDB-client 11.0+. > > Because of moves between packages MariaDB-Client conflicts with > MariaDB-server < 11.0.0. > > diff --git a/cmake/cpack_rpm.cmake b/cmake/cpack_rpm.cmake > index 638f6b2dabb..e88a5493eac 100644 > --- a/cmake/cpack_rpm.cmake > +++ b/cmake/cpack_rpm.cmake > @@ -189,7 +189,7 @@ FOREACH(SYM_COMPONENT Server Client) > SET(CPACK_RPM_${SYM}_PACKAGE_ARCHITECTURE "noarch") > SET(CPACK_RPM_${SYM}_USER_FILELIST ${ignored}) > STRING(TOLOWER ${SYM_COMPONENT} SYM_COMPONENT_LOWER) > - SET(CPACK_RPM_${SYM}_PACKAGE_REQUIRES "MariaDB-${SYM_COMPONENT_LOWER} >= 11.0.0") > + SETA(CPACK_RPM_${SYM}_PACKAGE_REQUIRES "MariaDB-${SYM_COMPONENT_LOWER} >= 11.0.0") why? > SETA(CPACK_RPM_${SYM_COMPONENT_LOWER}_PACKAGE_RECOMMENDS "MariaDB-${SYM}") > ENDFOREACH() > > @@ -205,10 +205,10 @@ SETA(CPACK_RPM_client_PACKAGE_PROVIDES > "mysql-client" > "mytop") > SETA(CPACK_RPM_client_PACKAGE_CONFLICTS > - "MariaDB-server < 10.6.0") > + "MariaDB-server < 11.0.0") because of resolve-stack-dump, ok > SETA(CPACK_RPM_common_PACKAGE_CONFLICTS > - "MariaDB-server < 10.6.1") > + "MariaDB-server < 11.0.0") why? > SETA(CPACK_RPM_devel_PACKAGE_OBSOLETES > "MySQL-devel") Regards, Sergei VP of MariaDB Server Engineering and security(a)mariadb.org

2 1

[Maria-developers] MDEV-30218: On what should be passed as 'rows' to filtering code.
by Sergey Petrunia 01 Feb '23

01 Feb '23

Hello Igor, So I'm looking at the bb-10.4-release-igor tree with your latest patch: commit c4fa6c3c4eaacefd9bf50b5c88c2c72474a15bc5 (HEAD -> bb-10.4-release-igor, origin/bb-10.4-release-igor) Author: Igor Babaev <igor(a)askmonty.org> Date: Tue Jan 31 13:14:53 2023 -0800 MDEV-30218 Incorrect optimization for rowid_filtering Correction over the last patch for this MDEV. I modify the code to add a printout: diff --git a/sql/sql_select.cc b/sql/sql_select.cc index 4bdfb659513..2cd4495bc03 100644 --- a/sql/sql_select.cc +++ b/sql/sql_select.cc @@ -7965,6 +7965,8 @@ best_access_path(JOIN *join, table->best_range_rowid_filter_for_partial_join(start_key->key, rows, access_cost_factor); + trace_access_idx.add("ROWS_FOR_FILTER", rows); + trace_access_idx.add("REAL_ROWS_FOR_FILTER", tmp); if (filter) { tmp-= filter->get_adjusted_gain(rows) - filter->get_cmp_gain(rows); then, I'm trying this example: create table t1 (a int, b int); insert into t1 select seq, seq from seq_1_to_10000; create table t2 (a int, key(a)); insert into t2 select seq from seq_1_to_10000; insert into t2 select * from t2; insert into t2 select * from t2; analyze table t2 persistent for all; create table t3 ( pk int primary key, a int, b int, unique key(a), key(b) ); insert into t3 select seq, seq, seq from seq_1_to_100000; set optimizer_trace=1; explain select straight_join * from t1, t2, t3 where t2.a=t1.a and t3.a=t1.b and t3.b < 5000; +------+-------------+-------+---------------+---------------+------+---------+---------+--------+---------------------------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +------+-------------+-------+---------------+---------------+------+---------+---------+--------+---------------------------------+ | 1 | SIMPLE | t1 | ALL | NULL | NULL | NULL | NULL | 10157 | Using where | | 1 | SIMPLE | t2 | ref | a | a | 5 | j1.t1.a | 4 | Using index | | 1 | SIMPLE | t3 | eq_ref|filter | a,b | a|b | 5|5 | j1.t1.b | 1 (5%) | Using where; Using rowid filter | +------+-------------+-------+---------------+---------------+------+---------+---------+--------+---------------------------------+ good so far. select * from information_schema.optimizer_trace; shows this (note the added printouts in caps): { "plan_prefix": ["t1", "t2"], "table": "t3", "best_access_path": { "considered_access_paths": [ { "access_type": "eq_ref", "index": "a", "ROWS_FOR_FILTER": 40628, "REAL_ROWS_FOR_FILTER": 10157, "rowid_filter_key": "b", "rows": 1, "cost": 9498, "chosen": true }, Let's run the query and check what really happens: ANALYZE FORMAT=JSON select straight_join * from t1, t2, t3 where t2.a=t1.a and t3.a=t1.b and t3.b < 5000; shows: ... "table": { "table_name": "t3", "access_type": "eq_ref", "key": "a", ... "rowid_filter": { ... "r_lookups": 10000, ... }, "r_loops": 40000, ... So, ROWS_FOR_FILTER=40K REAL_ROWS_FOR_FILTER=10K which of the numbers is a closer match here for the observed value of r_lookups= 10K ? BR Sergei -- Sergei Petrunia, Software Developer MariaDB Corporation | Skype: sergefp | Blog: http://petrunia.net

1 0

[Maria-developers] Unable to reset my mariadb root password.
by sanaboyina sarvani 01 Feb '23

01 Feb '23

Hello, I have just installed mariadb in my macos system. After installation it asks for the root password which I have not set yet. Tried to reset the root password but that did not work. So, how to know my mariadb root password? Is it possible to reset the password? Thank you. Regards

1 0