- developers - lists.mariadb.org

Re: [Maria-developers] Status of Windows build
by Kristian Nielsen 15 Jan '10

15 Jan '10

[moving the discussion to maria-developers@] Timour Katchaounov <timour(a)askmonty.org> writes: > Sergey, >> 1. buy VS Enterprise Edition (or Standard Edition, I wasn't able to figure out >> what exactly do we need to make builds. They don't offer evaluation versions of >> Standard, only Enterprise). >> >> 2. make the build script support building with VS Community Edition (possible >> but will take some time). > > No, we should not build with VS Community Edition. AFAIK, the free edition's > compiler produces much less optimized code than the payed for editions. I don't > think it is serious on our side to save money by not buying one VS license. On the other hand, the advantage of building with the community edition (I think we really mean the express / no-cost edition right?) is that it makes it easier for people to build themselves (and get the same result as our release builds). This is IMO important, we are an Open Source project after all. So it would be interesting to learn if there really is a speed difference here. Do you have some background information on this? Or do we need to run benchmarks to know for sure? > One more alternative would be to create a Windows virtual machine, and install > the evaluation version in the VM. Then once there is a copy, one can always > roll-back to the "fresh" installation. > > Given what is at stake, I'd much prefer that we have a license of VS. As I said above, I would prefer using the no-cost version, unless there is a real reason not to. Just my 2 cents, - Kristian.

5 7

[Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 2798: Fix for compiler warnings on windows
by noreply＠launchpad.net 14 Jan '10

14 Jan '10

------------------------------------------------------------ revno: 2798 committer: Michael Widenius <monty(a)askmonty.org> branch nick: maria-5.1 timestamp: Thu 2010-01-14 18:51:00 +0200 message: Fix for compiler warnings on windows Fix wrong cast of time() modified: include/my_pthread.h sql/handler.h storage/maria/ma_check.c storage/maria/ma_create.c storage/myisam/mi_check.c storage/myisam/mi_create.c storage/xtradb/handler/ha_innodb.cc -- 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] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (monty:2798)
by Michael Widenius 14 Jan '10

14 Jan '10

#At lp:maria based on revid:monty@askmonty.org-20100114163241-cvhq8ezp8psqy1aa 2798 Michael Widenius 2010-01-14 Fix for compiler warnings on windows Fix wrong cast of time() modified: include/my_pthread.h sql/handler.h storage/maria/ma_check.c storage/maria/ma_create.c storage/myisam/mi_check.c storage/myisam/mi_create.c storage/xtradb/handler/ha_innodb.cc per-file messages: include/my_pthread.h Safety fix that also removes compiler warnings sql/handler.h Changed timestamp columns to be of type time_t storage/maria/ma_check.c Removed wrong cast storage/maria/ma_create.c Removed wrong cast storage/myisam/mi_check.c Removed wrong cast storage/myisam/mi_create.c Removed wrong cast storage/xtradb/handler/ha_innodb.cc Removed compiler warning on windows === modified file 'include/my_pthread.h' --- a/include/my_pthread.h 2009-06-30 12:01:29 +0000 +++ b/include/my_pthread.h 2010-01-14 16:51:00 +0000 @@ -543,9 +543,9 @@ void safe_mutex_free_deadlock_data(safe_ #else #define my_pthread_mutex_init(A,B,C,D) pthread_mutex_init((A),(B)) #define my_pthread_mutex_lock(A,B) pthread_mutex_lock(A) -#define safe_mutex_assert_owner(mp) -#define safe_mutex_assert_not_owner(mp) -#define safe_mutex_free_deadlock_data(mp) +#define safe_mutex_assert_owner(mp) do {} while(0) +#define safe_mutex_assert_not_owner(mp) do {} while(0) +#define safe_mutex_free_deadlock_data(mp) do {} while(0) #endif /* SAFE_MUTEX */ #if defined(MY_PTHREAD_FASTMUTEX) && !defined(SAFE_MUTEX) === modified file 'sql/handler.h' --- a/sql/handler.h 2010-01-04 13:12:53 +0000 +++ b/sql/handler.h 2010-01-14 16:51:00 +0000 @@ -891,9 +891,9 @@ typedef struct { ulonglong delete_length; ha_rows records; ulong mean_rec_length; - ulong create_time; - ulong check_time; - ulong update_time; + time_t create_time; + time_t check_time; + time_t update_time; ulonglong check_sum; } PARTITION_INFO; @@ -1060,9 +1060,9 @@ public: ha_rows records; ha_rows deleted; /* Deleted records */ ulong mean_rec_length; /* physical reclength */ - ulong create_time; /* When table was created */ - ulong check_time; - ulong update_time; + time_t create_time; /* When table was created */ + time_t check_time; + time_t update_time; uint block_size; /* index block size */ ha_statistics(): === modified file 'storage/maria/ma_check.c' --- a/storage/maria/ma_check.c 2009-11-29 23:08:56 +0000 +++ b/storage/maria/ma_check.c 2010-01-14 16:51:00 +0000 @@ -6018,7 +6018,7 @@ int maria_update_state_info(HA_CHECK *pa { if (update & UPDATE_TIME) { - share->state.check_time= (long) time((time_t*) 0); + share->state.check_time= time((time_t*) 0); if (!share->state.create_time) share->state.create_time= share->state.check_time; } === modified file 'storage/maria/ma_create.c' --- a/storage/maria/ma_create.c 2009-02-19 09:01:25 +0000 +++ b/storage/maria/ma_create.c 2010-01-14 16:51:00 +0000 @@ -772,7 +772,7 @@ int maria_create(const char *name, enum share.base.min_block_length= share.base.pack_reclength; if (! (flags & HA_DONT_TOUCH_DATA)) - share.state.create_time= (long) time((time_t*) 0); + share.state.create_time= time((time_t*) 0); pthread_mutex_lock(&THR_LOCK_maria); === modified file 'storage/myisam/mi_check.c' --- a/storage/myisam/mi_check.c 2009-12-03 11:34:11 +0000 +++ b/storage/myisam/mi_check.c 2010-01-14 16:51:00 +0000 @@ -4447,7 +4447,7 @@ int update_state_info(HA_CHECK *param, M { if (update & UPDATE_TIME) { - share->state.check_time= (long) time((time_t*) 0); + share->state.check_time= time((time_t*) 0); if (!share->state.create_time) share->state.create_time=share->state.check_time; } === modified file 'storage/myisam/mi_create.c' --- a/storage/myisam/mi_create.c 2009-10-15 21:38:29 +0000 +++ b/storage/myisam/mi_create.c 2010-01-14 16:51:00 +0000 @@ -575,7 +575,7 @@ int mi_create(const char *name,uint keys max(share.base.pack_reclength,MI_MIN_BLOCK_LENGTH) : MI_EXTEND_BLOCK_LENGTH; if (! (flags & HA_DONT_TOUCH_DATA)) - share.state.create_time= (long) time((time_t*) 0); + share.state.create_time= time((time_t*) 0); pthread_mutex_lock(&THR_LOCK_myisam); === modified file 'storage/xtradb/handler/ha_innodb.cc' --- a/storage/xtradb/handler/ha_innodb.cc 2009-12-03 11:34:11 +0000 +++ b/storage/xtradb/handler/ha_innodb.cc 2010-01-14 16:51:00 +0000 @@ -61,7 +61,9 @@ with this program; if not, write to the #pragma implementation // gcc: Class implementation #endif +#ifndef MYSQL_SERVER #define MYSQL_SERVER +#endif #include <mysql_priv.h> #ifdef MYSQL_SERVER

1 0

[Maria-developers] Merge directive: Compilation error and warning fixes on Win32 & time fix.
by Alex Budovski 14 Jan '10

14 Jan '10

See log messages for details. - Alex

2 1

[Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 2797: Fixed compile error on windows.
by noreply＠launchpad.net 14 Jan '10

14 Jan '10

------------------------------------------------------------ revno: 2797 committer: Michael Widenius <monty(a)askmonty.org> branch nick: maria-5.1 timestamp: Thu 2010-01-14 18:32:41 +0200 message: Fixed compile error on windows. modified: sql/set_var.cc -- 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] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (monty:2797)
by Michael Widenius 14 Jan '10

14 Jan '10

#At lp:maria based on revid:monty@askmonty.org-20100112173111-yj9651464x04ggba 2797 Michael Widenius 2010-01-14 Fixed compile error on windows. modified: sql/set_var.cc === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-12-03 11:19:05 +0000 +++ b/sql/set_var.cc 2010-01-14 16:32:41 +0000 @@ -58,6 +58,9 @@ #include <my_getopt.h> #include <thr_alarm.h> #include <myisam.h> +#ifdef WITH_MARIA_STORAGE_ENGINE +#include <maria.h> +#endif #include <my_dir.h> #include <waiting_threads.h> #include "events.h"

1 0

[Maria-developers] Progress (by Bothorsen): Provide an " ORDER BY FIRST_JOIN.column" feature (59)
by worklog-noreply＠askmonty.org 14 Jan '10

14 Jan '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Provide an "ORDER BY FIRST_JOIN.column" feature CREATION DATE..: Thu, 22 Oct 2009, 12:33 SUPERVISOR.....: Bothorsen IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-RawIdeaBin TASK ID........: 59 (http://askmonty.org/worklog/?tid=59) VERSION........: WorkLog-3.4 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 5 ESTIMATE.......: 60 (hours remain) ORIG. ESTIMATE.: 60 PROGRESS NOTES: -=-=(Bothorsen - Thu, 14 Jan 2010, 15:24)=-=- Estimated by Igor to be 10 to 12 days after estimation. Worked 5 hours and estimate 60 hours remain (original estimate increased by 65 hours). DESCRIPTION: Question: How much effort would be required to provide an "ORDER BY FIRST_JOIN.column" feature? We often do self-join queries and want to order them by index, i.e.: create table t1 (a int, b int, unique key 'by_a_b' (a, b))); select * from t1 x inner join t1 y on (x.b = y.b) where x.a=1 and y.a=2 order by (x|y).b desc limit 10; but it is only ordered by index if you order by the first table in the optimized join order, which could be either x or y depending on which is more selective. we don't want to force the join order because often one is far more selective than the other, but the only way to know what table to order by then is to explain the query first. if there was some way to tell mysql to just order by the first join in the optimized order, that would help us. sometimes we do this across tables too. both of these can be solved by rewriting the query to say "using" instead of "on" in which case we don't have to specify the table name of the column, just "order by b desc". but, we also want to be able to do exclusions, in which case an "on" is required and therefore we run into ambiguous column names: Monty answered: MySQL has an optimization where it knows that if x.b = y.b is used then it can replace x.b with y.b and y.b with x.b in the WHERE part MySQL however doesn't do it for the ORDER BY part and I don't think that should be very hard to do. Question continues: create table t2 (c int, b int, unique key 'by_c_b' (c, b))); select * from t1 x inner join t1 y using (b) left join t2 on (t2.c = 3 and t1.b = t2.b) where x.a=1 and y.a=2 and t2.c is null order by (x| y).b desc limit 10; if we were ordering ascending, i think we could just leave off the order by entirely in this case and it would happen to work since it's reading in index order. but the combination of requiring an "on" and descending sort leaves us unable to use these tricks. if we had either an "ORDER BY FIRST_JOIN.column" or some way to tell the server that 'b' is in fact joined as being equivalent across tables (except where it may be null from left joining) and that we shouldn't have to specify the table name at all, that would save us having to figure out the table from the explain. Monty answers: It may be that the eq-replacment we have would solve this. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): Provide key cache statistics (58)
by worklog-noreply＠askmonty.org 14 Jan '10

14 Jan '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Provide key cache statistics CREATION DATE..: Thu, 22 Oct 2009, 12:28 SUPERVISOR.....: Bothorsen IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-RawIdeaBin TASK ID........: 58 (http://askmonty.org/worklog/?tid=58) VERSION........: WorkLog-3.4 STATUS.........: Cancelled PRIORITY.......: 60 WORKED HOURS...: 0 ESTIMATE.......: 0 (hours remain) ORIG. ESTIMATE.: 0 PROGRESS NOTES: -=-=(Guest - Thu, 14 Jan 2010, 15:15)=-=- Status updated. --- /tmp/wklog.58.old.19489 2010-01-14 13:15:35.000000000 +0000 +++ /tmp/wklog.58.new.19489 2010-01-14 13:15:35.000000000 +0000 @@ -1 +1 @@ -Un-Assigned +Cancelled -=-=(Monty - Fri, 27 Nov 2009, 12:36)=-=- High Level Description modified. --- /tmp/wklog.58.old.15108 2009-11-27 10:36:13.000000000 +0000 +++ /tmp/wklog.58.new.15108 2009-11-27 10:36:13.000000000 +0000 @@ -1,8 +1,7 @@ -Provide the key cache -statistics available for the default cache (key blocks used, -unflushed, etc.) for all named caches. +Provide the same key cache statistics for all MyISAM key caches as we provide +for the default key cache. -Monty answers in an email: +Background: Monty answers in an email to customer: This is a much simpler task and something that is important to get done. We would have to introduce a 'show keycache statistics' command @@ -11,3 +10,44 @@ The raw coding is probably 1-2 days, but on top if this we need testing, a test environment and building of biniaries for you. Henrik/Bo and Igor will come back to you with a more exact estimate. + +----------------- +Currently we have: + +show status like "key%"; ++------------------------+-------+ +| Variable_name | Value | ++------------------------+-------+ +| Key_blocks_not_flushed | 0 | +| Key_blocks_unused | 13389 | +| Key_blocks_used | 7 | +| Key_read_requests | 22 | +| Key_reads | 7 | +| Key_write_requests | 0 | +| Key_writes | 0 | ++------------------------+-------+ + +Sergei suggested we introduce for each new key cache a set of variables prefixed +with the keycache name: + +show status like "keycache1.%"; +Which should show: + ++----------------------------------+-------+ +| Variable_name | Value | ++----------------------------------+-------+ +| keycache1.Key_blocks_not_flushed | 0 | +| keycache1.Key_blocks_unused | 13389 | +| keycache1.Key_blocks_used | 7 | +| keycache1.Key_read_requests | 22 | +| keycache1.Key_reads | 7 | +| keycache1.Key_write_requests | 0 | +| keycache1.Key_writes | 0 | ++----------------------------------+-------+ + +The task would thus to be to automatically introduce new variables when we +create a new key cache and automatically remove these variables when the cache +is deleted. + +The other option would be to add an information schema where we populate the +information schema with data from all existing key caches. DESCRIPTION: Provide the same key cache statistics for all MyISAM key caches as we provide for the default key cache. Background: Monty answers in an email to customer: This is a much simpler task and something that is important to get done. We would have to introduce a 'show keycache statistics' command that would iterate over all keycaches and provide the statistics. The raw coding is probably 1-2 days, but on top if this we need testing, a test environment and building of biniaries for you. Henrik/Bo and Igor will come back to you with a more exact estimate. ----------------- Currently we have: show status like "key%"; +------------------------+-------+ | Variable_name | Value | +------------------------+-------+ | Key_blocks_not_flushed | 0 | | Key_blocks_unused | 13389 | | Key_blocks_used | 7 | | Key_read_requests | 22 | | Key_reads | 7 | | Key_write_requests | 0 | | Key_writes | 0 | +------------------------+-------+ Sergei suggested we introduce for each new key cache a set of variables prefixed with the keycache name: show status like "keycache1.%"; Which should show: +----------------------------------+-------+ | Variable_name | Value | +----------------------------------+-------+ | keycache1.Key_blocks_not_flushed | 0 | | keycache1.Key_blocks_unused | 13389 | | keycache1.Key_blocks_used | 7 | | keycache1.Key_read_requests | 22 | | keycache1.Key_reads | 7 | | keycache1.Key_write_requests | 0 | | keycache1.Key_writes | 0 | +----------------------------------+-------+ The task would thus to be to automatically introduce new variables when we create a new key cache and automatically remove these variables when the cache is deleted. The other option would be to add an information schema where we populate the information schema with data from all existing key caches. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Progress (by Knielsen): Enable the use of libstdc++ in MariaDB (63)
by worklog-noreply＠askmonty.org 14 Jan '10

14 Jan '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Enable the use of libstdc++ in MariaDB CREATION DATE..: Wed, 11 Nov 2009, 13:19 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-BackLog TASK ID........: 63 (http://askmonty.org/worklog/?tid=63) VERSION........: Server-5.2 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 4 ESTIMATE.......: 10 (hours remain) ORIG. ESTIMATE.: 10 PROGRESS NOTES: -=-=(Knielsen - Thu, 14 Jan 2010, 13:46)=-=- Research and updated description Worked 4 hours and estimate 10 hours remain (original estimate increased by 14 hours). -=-=(Knielsen - Thu, 14 Jan 2010, 13:45)=-=- High-Level Specification modified. --- /tmp/wklog.63.old.13967 2010-01-14 11:45:49.000000000 +0000 +++ /tmp/wklog.63.new.13967 2010-01-14 11:45:49.000000000 +0000 @@ -1 +1,37 @@ +I did some investigation into this. + +The simple way to do this is to simply use g++ to link C++ objects. So this +issue is really restricted to GCC compilation where we by default prefer to +link with gcc even for C++ code. So this means building with + + CXX=g++ + +The consequences of doing this for the binaries is the addition of two +additional run-time .so dependencies: libstdc++.so and libgcc_s.so. + +It still needs to be investigated if these additional dependencies are a +problem for binary tarball packages, or if the ABI for those libraries are now +as stable as libc.so. + +The libgcc_s.so is needed as a dependency to support exceptions between +different object files, as they need to use the same code for stack unwinding. +libstdc++.so is of course needed for access to C++ runtime. + +I researched into the possibility to instead link only specific plugins with +g++, and continue to link the rest of the server with gcc. Unfortunately, this +seems really hard to do in a proper way due to the way autotools works. At +configure time, a script ./libtool is created with hardcoded compiler commands +derived from $CC and $CXX. This script is then used to do the actual linking +in Makefiles generated by Automake. I thus did not find a way to change the +linker command on a per-makefile basis, as libtool is global to the project. + +One option would be to use separate configure.in for plugins, but this is +quite an intrusive change. + +My conclusion is that the best way is to start using g++ for linking the +entire server. This is no problem for binaries made for a specific +distribution (Ubuntu, Debian, Centos), where the dependencies are handled by +the package manager. If it is a big problem for binary tarball releases, at +worst we can build multiple binary tarball releases for the different library +versions we need to support. -=-=(Guest - Tue, 12 Jan 2010, 16:26)=-=- Version updated. --- /tmp/wklog.63.old.19522 2010-01-12 16:26:23.000000000 +0200 +++ /tmp/wklog.63.new.19522 2010-01-12 16:26:23.000000000 +0200 @@ -1 +1 @@ -Connector/.NET-5.2 +Server-5.2 -=-=(Guest - Tue, 12 Jan 2010, 16:26)=-=- Category updated. --- /tmp/wklog.63.old.19506 2010-01-12 16:26:15.000000000 +0200 +++ /tmp/wklog.63.new.19506 2010-01-12 16:26:15.000000000 +0200 @@ -1 +1 @@ -Server-RawIdeaBin +Server-BackLog DESCRIPTION: Enable the use of libstdc++ in MariaDB. As time goes on, more and more plugins and external code/library will need linking to libstdc++ for stuff it uses. I have already seen this happen several times, with extra work needed to integrate things properly. It would be nice to have a general solution for this so that it is not necessary to spend time on individual solutions in each case. It also needs to be considered what the impact of this will be for the server in terms of binary compatibility, performance etc. I think it should be mostly ok, except that it might introduce a problem for bintar packages with an external dependency on libstdc++. HIGH-LEVEL SPECIFICATION: I did some investigation into this. The simple way to do this is to simply use g++ to link C++ objects. So this issue is really restricted to GCC compilation where we by default prefer to link with gcc even for C++ code. So this means building with CXX=g++ The consequences of doing this for the binaries is the addition of two additional run-time .so dependencies: libstdc++.so and libgcc_s.so. It still needs to be investigated if these additional dependencies are a problem for binary tarball packages, or if the ABI for those libraries are now as stable as libc.so. The libgcc_s.so is needed as a dependency to support exceptions between different object files, as they need to use the same code for stack unwinding. libstdc++.so is of course needed for access to C++ runtime. I researched into the possibility to instead link only specific plugins with g++, and continue to link the rest of the server with gcc. Unfortunately, this seems really hard to do in a proper way due to the way autotools works. At configure time, a script ./libtool is created with hardcoded compiler commands derived from $CC and $CXX. This script is then used to do the actual linking in Makefiles generated by Automake. I thus did not find a way to change the linker command on a per-makefile basis, as libtool is global to the project. One option would be to use separate configure.in for plugins, but this is quite an intrusive change. My conclusion is that the best way is to start using g++ for linking the entire server. This is no problem for binaries made for a specific distribution (Ubuntu, Debian, Centos), where the dependencies are handled by the package manager. If it is a big problem for binary tarball releases, at worst we can build multiple binary tarball releases for the different library versions we need to support. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Progress (by Knielsen): Enable the use of libstdc++ in MariaDB (63)
by worklog-noreply＠askmonty.org 14 Jan '10

14 Jan '10

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Enable the use of libstdc++ in MariaDB CREATION DATE..: Wed, 11 Nov 2009, 13:19 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: CATEGORY.......: Server-BackLog TASK ID........: 63 (http://askmonty.org/worklog/?tid=63) VERSION........: Server-5.2 STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 4 ESTIMATE.......: 10 (hours remain) ORIG. ESTIMATE.: 10 PROGRESS NOTES: -=-=(Knielsen - Thu, 14 Jan 2010, 13:46)=-=- Research and updated description Worked 4 hours and estimate 10 hours remain (original estimate increased by 14 hours). -=-=(Knielsen - Thu, 14 Jan 2010, 13:45)=-=- High-Level Specification modified. --- /tmp/wklog.63.old.13967 2010-01-14 11:45:49.000000000 +0000 +++ /tmp/wklog.63.new.13967 2010-01-14 11:45:49.000000000 +0000 @@ -1 +1,37 @@ +I did some investigation into this. + +The simple way to do this is to simply use g++ to link C++ objects. So this +issue is really restricted to GCC compilation where we by default prefer to +link with gcc even for C++ code. So this means building with + + CXX=g++ + +The consequences of doing this for the binaries is the addition of two +additional run-time .so dependencies: libstdc++.so and libgcc_s.so. + +It still needs to be investigated if these additional dependencies are a +problem for binary tarball packages, or if the ABI for those libraries are now +as stable as libc.so. + +The libgcc_s.so is needed as a dependency to support exceptions between +different object files, as they need to use the same code for stack unwinding. +libstdc++.so is of course needed for access to C++ runtime. + +I researched into the possibility to instead link only specific plugins with +g++, and continue to link the rest of the server with gcc. Unfortunately, this +seems really hard to do in a proper way due to the way autotools works. At +configure time, a script ./libtool is created with hardcoded compiler commands +derived from $CC and $CXX. This script is then used to do the actual linking +in Makefiles generated by Automake. I thus did not find a way to change the +linker command on a per-makefile basis, as libtool is global to the project. + +One option would be to use separate configure.in for plugins, but this is +quite an intrusive change. + +My conclusion is that the best way is to start using g++ for linking the +entire server. This is no problem for binaries made for a specific +distribution (Ubuntu, Debian, Centos), where the dependencies are handled by +the package manager. If it is a big problem for binary tarball releases, at +worst we can build multiple binary tarball releases for the different library +versions we need to support. -=-=(Guest - Tue, 12 Jan 2010, 16:26)=-=- Version updated. --- /tmp/wklog.63.old.19522 2010-01-12 16:26:23.000000000 +0200 +++ /tmp/wklog.63.new.19522 2010-01-12 16:26:23.000000000 +0200 @@ -1 +1 @@ -Connector/.NET-5.2 +Server-5.2 -=-=(Guest - Tue, 12 Jan 2010, 16:26)=-=- Category updated. --- /tmp/wklog.63.old.19506 2010-01-12 16:26:15.000000000 +0200 +++ /tmp/wklog.63.new.19506 2010-01-12 16:26:15.000000000 +0200 @@ -1 +1 @@ -Server-RawIdeaBin +Server-BackLog DESCRIPTION: Enable the use of libstdc++ in MariaDB. As time goes on, more and more plugins and external code/library will need linking to libstdc++ for stuff it uses. I have already seen this happen several times, with extra work needed to integrate things properly. It would be nice to have a general solution for this so that it is not necessary to spend time on individual solutions in each case. It also needs to be considered what the impact of this will be for the server in terms of binary compatibility, performance etc. I think it should be mostly ok, except that it might introduce a problem for bintar packages with an external dependency on libstdc++. HIGH-LEVEL SPECIFICATION: I did some investigation into this. The simple way to do this is to simply use g++ to link C++ objects. So this issue is really restricted to GCC compilation where we by default prefer to link with gcc even for C++ code. So this means building with CXX=g++ The consequences of doing this for the binaries is the addition of two additional run-time .so dependencies: libstdc++.so and libgcc_s.so. It still needs to be investigated if these additional dependencies are a problem for binary tarball packages, or if the ABI for those libraries are now as stable as libc.so. The libgcc_s.so is needed as a dependency to support exceptions between different object files, as they need to use the same code for stack unwinding. libstdc++.so is of course needed for access to C++ runtime. I researched into the possibility to instead link only specific plugins with g++, and continue to link the rest of the server with gcc. Unfortunately, this seems really hard to do in a proper way due to the way autotools works. At configure time, a script ./libtool is created with hardcoded compiler commands derived from $CC and $CXX. This script is then used to do the actual linking in Makefiles generated by Automake. I thus did not find a way to change the linker command on a per-makefile basis, as libtool is global to the project. One option would be to use separate configure.in for plugins, but this is quite an intrusive change. My conclusion is that the best way is to start using g++ for linking the entire server. This is no problem for binaries made for a specific distribution (Ubuntu, Debian, Centos), where the dependencies are handled by the package manager. If it is a big problem for binary tarball releases, at worst we can build multiple binary tarball releases for the different library versions we need to support. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0