OSDN Git Service
Kyotaro Horiguchi [Wed, 28 Oct 2020 06:50:58 +0000 (15:50 +0900)]
Make HashJoin hint more coercive.
Even with HashJoin hint, hash joins may be rejected by planner if hash
table for the inner-rel is estimated too large. Make HashJoin hint
more coercive by increasing hash_mem_multiplier.
Kyotaro Horiguchi [Thu, 29 Oct 2020 12:02:28 +0000 (21:02 +0900)]
Change copyright year of docs to 2020
Kyotaro Horiguchi [Wed, 28 Oct 2020 13:58:48 +0000 (22:58 +0900)]
Stabilize regression test.
I was annoyed by frequent nonsubstantial change of title line length
of explain output. This commit makes maskout.sh replace the title
lines with fixed content.
Kyotaro Horiguchi [Wed, 28 Oct 2020 09:03:05 +0000 (18:03 +0900)]
Follow a behavioral change in core.
Commit
894f5dea76 has changed the core's behavior of naming indexes.
Follow the changes. This is regression-result changes only.
Kyotaro Horiguchi [Thu, 29 Oct 2020 04:54:40 +0000 (13:54 +0900)]
Update core.c and make_join_rel.c.
From this version core.c and make_join_rel.c are automatically
generated from core code. Adjust them to auto-generated content.
Kyotaro Horiguchi [Thu, 29 Oct 2020 04:45:37 +0000 (13:45 +0900)]
New tool to generate source files for copied functions
Previously core.c and make_join_rel.c are maintained by hand. This
tool generates the files from corresponding core source files.
Kyotaro Horiguchi [Thu, 29 Oct 2020 00:08:47 +0000 (09:08 +0900)]
Add *.bc to .gitignore
Added *.bc and *~ to .gitignore
Kyotaro Horiguchi [Wed, 28 Oct 2020 09:00:28 +0000 (18:00 +0900)]
Follow the semantic change of pg_list part 2
Commit
7ed93ce forgot to fix another place where list_delete_cell() is
used. Fix it.
Kyotaro Horiguchi [Thu, 6 Aug 2020 11:18:05 +0000 (20:18 +0900)]
Support PostgreSQL 13.
- Follow change of some planner functions.
- Follow a change of the semantics of List.
- Take in changes in copied core code.
- Adjust regression tests.
Amost all of the differences are come from aliasing policy, and some
come from a new GUC parameter.
Kyotaro Horiguchi [Wed, 5 Aug 2020 07:30:25 +0000 (16:30 +0900)]
Version 1.3.6.
Fixed some bugs.
sherlockcpp [Sun, 5 Apr 2020 13:18:31 +0000 (21:18 +0800)]
Add description of *ScanRegExp hints in the document.
English version of hint list is missing regexp versions of scan
hints. Add them.
Kyotaro Horiguchi [Tue, 14 Jul 2020 08:17:19 +0000 (17:17 +0900)]
Mask non-significant parts of regression output
To stabilize test, mask more of non-significant parts of EXPLAIN output.
Daniil Anisimov [Mon, 16 Mar 2020 08:11:12 +0000 (15:11 +0700)]
Fix segfault when use Set, Rows and Parallel hints together
Yang Rong [Thu, 21 May 2020 11:38:57 +0000 (19:38 +0800)]
Fix some spelling errors in comments
egashira.yusuke [Fri, 26 Jun 2020 11:09:51 +0000 (20:09 +0900)]
Mark hint table as a configuration table.
The table hint_plan.hints is a configuration table of
pg_hint_plan. Mark it as such, as well as its sequence table.
Back patch to pg_hint_plan96.
Kyotaro Horiguchi [Thu, 20 Feb 2020 12:24:14 +0000 (21:24 +0900)]
Preparation for branch PG12
Change version to 12-1.3.5. Adjust SPEC file.
Kyotaro Horiguchi [Thu, 20 Feb 2020 11:53:24 +0000 (20:53 +0900)]
Support PG12.
As the prepparation for making PG12 branch, change PG version to 12
and adjust for the latest PG12.
Kyotaro Horiguchi [Thu, 20 Feb 2020 11:19:44 +0000 (20:19 +0900)]
Change copyright year to 2020.
Kyotaro Horiguchi [Mon, 17 Feb 2020 10:46:28 +0000 (19:46 +0900)]
Restore current hint state when returned from non-hinted query planning.
If no hint is given for the current level query, pg_hint_plan_planner
calls the next level of planner after erasing the
current_hint_state. But it forgot to restore the state before the
planning of the rest part of the current-level query. It is
(a-kind-of) broken by the commit
d422966 but overlooked as an
inevitable side-effect of the fix. Get back the behavior by restoring
current_hint_state after returned from the lower level planner with
unhinted query.
Issue: https://github.com/ossc-db/pg_hint_plan/issues/30
Reported-by: higuchi-daisuke
Kyotaro Horiguchi [Mon, 17 Feb 2020 07:19:48 +0000 (16:19 +0900)]
Use proper type for non-integer GUC options.
pg_hint_plan internally preserved numeric options into integer
variable regardless of their types, which leads to unintentional
change of the values. Handle double values in the proper way.
Kyotaro Horiguchi [Mon, 17 Feb 2020 05:53:27 +0000 (14:53 +0900)]
Fix Rows hint parsing
This is a long standing bug that Rows hint with no parameter causees a
crash. Fixed the Rows hint parser.
Kyotaro Horiguchi [Mon, 17 Feb 2020 04:15:39 +0000 (13:15 +0900)]
Give correct query location to generate_normalized_query.
There was a thinko about query_loc of generate_normalized_query which
causes a crash. For example a query is following a sequence of
semicolons (by \; in psql input) longer than the length after the last
placeholder to the end of the query, the function causes assertion
failure. Anyway prepared statements lose that value so we just pass 0
as the query_loc for consistent behavior.
Kyotaro Horiguchi [Fri, 14 Feb 2020 04:44:27 +0000 (13:44 +0900)]
Fix behavior of nested planning
Issue: https://github.com/ossc-db/pg_hint_plan/issues/39
When planning is nested by function call, the hints in the function
should affect the corresponding query. Maybe
9599067 introduced
that. Fix the nested planning behavior.
Kyotaro Horiguchi [Fri, 14 Feb 2020 03:34:55 +0000 (12:34 +0900)]
Fix crash bug caused by plancache invalidation
https://github.com/ossc-db/pg_hint_plan/issues/41
After plancache is invalidated then revaliated, get_query_string
accesses query_list of invalid plansource then crash. Ignore invalid
plancache and get the correct hint string and Query node at the next
planning time during revalidation.
On the way fixing this, a bug related to planner reentrance is
fixed. That fix causes behavioral change for nested
planning. Previously outer-level hint (wrongly) overrides inner-level
query but currenlty outer-level hint no longer affects inner-level
query.
LittleWuCoding [Sat, 12 Jan 2019 10:11:04 +0000 (18:11 +0800)]
Fix compiling error on Solaris
The 'LDFLAGS+=-Wl,--build-id' option is for rpm building in Linux like
OS. The SunOS is not native support the option
'LDFLAGS+=-Wl,--build-id' . So, we disable this option in SunOS while
compiling.
Committed a bit tweaked version from the orignal.
Kyotaro Horiguchi [Thu, 13 Feb 2020 05:57:07 +0000 (14:57 +0900)]
Fix the previous commit.
The previous commit left a line to be removed. Remove it.
Kyotaro Horiguchi [Thu, 13 Feb 2020 05:34:01 +0000 (14:34 +0900)]
Do null-check of error_context_stack before use
The variable is accessed without null checking when planner is called
in the context of pl/pgsql. Fix it. Back-patched up to pg_hint_plan94.
Reporter: egashira-yusuke
Issue: https://github.com/ossc-db/pg_hint_plan/issues/42
Original-Author: sun-wei <
30999182+Dagouhan@users.noreply.github.com>
Original-patch: https://github.com/ossc-db/pg_hint_plan/pull/45/commits/
8fe092ef73a9af6cfb41e20717d7cb35aa74fce2
Kyotaro Horiguchi [Tue, 29 Oct 2019 09:49:22 +0000 (18:49 +0900)]
Fix regression tests.
Fix regression tests to work for PG12.
Kyotaro Horiguchi [Tue, 29 Oct 2019 09:47:17 +0000 (18:47 +0900)]
Follow the change in PG12.
Commit
428b260f87 breaks the prior assumptions by adding one more
planning cycle for updates on partitioned tables. Follow the change.
Kyotaro Horiguchi [Tue, 29 Oct 2019 09:24:38 +0000 (18:24 +0900)]
Jignesh Shah [Sun, 2 Jun 2019 20:23:05 +0000 (20:23 +0000)]
Fix definition of is_dummy_rel based on changes in community
LittleWuCoding [Wed, 12 Dec 2018 06:47:13 +0000 (14:47 +0800)]
fix spelling mistakes in file pg_hint_plan.html
Arthur Zakirov [Wed, 1 Aug 2018 16:27:23 +0000 (19:27 +0300)]
Use ParallelHint in ParallelHintCreate()
Kyotaro Horiguchi [Fri, 8 Mar 2019 04:58:52 +0000 (13:58 +0900)]
Silence some compilers
Some compilers seem to complain about the previous patch, that
prev_hint_str is used uninitialized. Actually it is not but move the
initialization code out of the PG_TRY() section so that such compilers
would be quiet.
Kyotaro Horiguchi [Tue, 26 Feb 2019 07:10:29 +0000 (16:10 +0900)]
Correctly handle planner nesting
pg_hint_plan assumed that plpgsql is the only source of nested planner
calls. Actually nested call can be made in any shapes. Most of the
cases doesn't harm but in a special case where pg_dbms_stats makes a
SPI call during query planning, that affects subsequent planner
work. Hints lose effect when pg_dbms_stats searches "locked
statistics" tables while planning the target query.
9.5 is not affected. Back patched back to 9.6.
Kyotaro Horiguchi [Tue, 26 Feb 2019 10:25:08 +0000 (19:25 +0900)]
Follow core's cobe/havor chages
Take in some changes took place in core code.
Commit
608b167f9f has largely changed the behavior around
CTE. Although I didn't confirmed that the all changes are valid as the
test for pg_hint_plan, just took in all plan differences.
Kyotaro Horiguchi [Thu, 17 Jan 2019 09:53:12 +0000 (18:53 +0900)]
Change version to 1.3.4.
Fixed a minor bug of 1.3.3.
Kyotaro Horiguchi [Thu, 17 Jan 2019 09:32:57 +0000 (18:32 +0900)]
Fix condition to regenerate gather paths
Core has additional condition to determine whether to generate gather
paths and pg_hint_plan forgot it. Fix it.
Kyotaro Horiguchi [Thu, 17 Jan 2019 03:33:00 +0000 (12:33 +0900)]
Change version to 1.3.3
Version bumped to 1.3.3. SPEC file is updated.
Kyotaro Horiguchi [Thu, 17 Jan 2019 07:37:15 +0000 (16:37 +0900)]
Fix reverse link of hint_list.html
pg_hint_plan.html and hint_list.html pointed to the Japanese version
of main document. Fixed it.
Kyotaro Horiguchi [Thu, 17 Jan 2019 03:29:47 +0000 (12:29 +0900)]
Fix documentation.
Fixed wrong or stale descriptions, and poor or broken styles of
the documentation.
Kyotaro Horiguchi [Thu, 17 Jan 2019 03:29:47 +0000 (12:29 +0900)]
Include all pg_hint_plan--*.sql files in installation
make install didn't install upgrade scripts. Fix it.
Kyotaro Horiguchi [Thu, 17 Jan 2019 03:11:17 +0000 (12:11 +0900)]
Change copyright year to 2019
Kyotaro Horiguchi [Thu, 17 Jan 2019 02:52:07 +0000 (11:52 +0900)]
Fix content of llvmjit package
The previous release is missing index.bc file and mislocating .bc file
in the llvmjit subpackage. The former is created at make install time
so use the standard make file to rpm-build.
Kyotaro Horiguchi [Wed, 16 Jan 2019 08:28:47 +0000 (17:28 +0900)]
Catch up with the latest master
Some refactoring of header files and commits
afb0d0712f and
1db5667bac
of core hit this. Catch up with them.
Kyotaro Horiguchi [Wed, 9 Jan 2019 00:35:41 +0000 (09:35 +0900)]
Call standrad_ProcessUtility when no hook is set
The last commit
d5ec243 forgot the case of no previous hook function
in the ProcessUtility hook function. Fix it.
Kyotaro Horiguchi [Mon, 7 Jan 2019 08:01:32 +0000 (17:01 +0900)]
Support prepared statements on extended protocol
However pg_hint_plan doesn't fully consider the extended protocol,
commit
c05bb31 accidentially broke the case where an analyzed prepared
statement is executed on extended protocol. This patch fixes that only
for the hints-in-comment case. Hint-table still doesn't work in the
case since query-normalization needs Query, which is not available in
planner_hook.
Kyotaro Horiguchi [Mon, 7 Jan 2019 07:40:21 +0000 (16:40 +0900)]
Remove an useless variable
commit
c05bb31 made the variable stmt_name useless. Remove it.
Kyotaro Horiguchi [Mon, 7 Jan 2019 09:57:53 +0000 (18:57 +0900)]
Adjust regression test
Core behavior has changed, adjust regression test.
Kyotaro Horiguchi [Wed, 5 Dec 2018 07:56:11 +0000 (16:56 +0900)]
Adjust regression test
Some regression test items need adjustment due to change of server
behavior. The way the regression test ignores unstable portions of
EXPLAIN output is changed along with that.
Kyotaro Horiguchi [Tue, 4 Dec 2018 04:19:57 +0000 (13:19 +0900)]
Fix for union-on-inheritance case
setup_hint_enforcement may pick up a wrong hint when multiple
subqueries access inheritance tables. This leads to failure to apply
hints on other than the first subquery.
Kyotaro Horiguchi [Tue, 4 Dec 2018 01:33:55 +0000 (10:33 +0900)]
Don't reset plpgsql nest level at every command end
Clean up code for global exit is forgetting about loops that run
hinted commands multiple times. This leads to pg_hint_plan's failure
to affect after the first execution of the inner commands.
The hint below worked only for the first time.
FOR outer IN EXECUTE 'outer-command' LOOP
FOR inner IN EXECUTE '/*+ hint */ inner-command' LOOP
<something>
END LOOP;
END LOOP;
Kyotaro Horiguchi [Tue, 13 Nov 2018 09:02:35 +0000 (18:02 +0900)]
Minor document fix
Just fix versions of PostgreSQL and OS.
Kyotaro Horiguchi [Tue, 13 Nov 2018 01:29:17 +0000 (10:29 +0900)]
Change target version to PG11 and fix the SPEC file
Change targetr version to PG11. SPEC file was changed so that genearte
a subpackage llvmjit, which installs a bitcode file needed for JIT
compilation.
Kyotaro Horiguchi [Tue, 13 Nov 2018 04:05:06 +0000 (13:05 +0900)]
Fix copyright year
Kyotaro Horiguchi [Tue, 13 Nov 2018 02:42:22 +0000 (11:42 +0900)]
Fix regression test following core's behavior
Some changes has been made on GUC default value and behavior. Follow
the change. No change was made on pg_hint_plan's behavior.
Kyotaro Horiguchi [Tue, 13 Nov 2018 04:37:54 +0000 (13:37 +0900)]
Update core.c and make_jion_rel.c following core
Apply changes ever made on core.
Kyotaro Horiguchi [Tue, 13 Nov 2018 06:59:57 +0000 (15:59 +0900)]
Change version to 1.3.2.
Kyotaro Horiguchi [Tue, 13 Nov 2018 03:43:10 +0000 (12:43 +0900)]
Fix debug output level control
Previously parse_messages and message_level were binded to the same
internal variable. Separate the two GUCs. On the way doing this some
mssages that did not honor the variable were fixed. Default behavior
is not changed but the the two variables works differently from the
previous version, perhaps closer to expected.
Kyotaro Horiguchi [Tue, 13 Nov 2018 00:27:48 +0000 (09:27 +0900)]
Fix bogus error handling
For somewhat reason set_config_int32_options does bogus error
handling. Howerver that error cannot be happen, fixed it.
Kyotaro Horiguchi [Tue, 13 Nov 2018 06:54:20 +0000 (15:54 +0900)]
Change version to 1.3.1.
1.3.0 has a crash bug in Describe message of extended protocol
path. If server receives the message after invalidation of
corresponding query cache, pg_hint_plan crashes.
Kyotaro Horiguchi [Fri, 8 Jun 2018 03:27:03 +0000 (12:27 +0900)]
Took in several improvements of pg_stat_statements
Following two commits are took in.
83f2061dd0 Teach contrib/pg_stat_statements to handle multi-statement commands better.
0bb51aa967 Improve parsetree representation of special functions such as CURRENT_DATE.
The following one makes change of documented behavior of pg_hint_plan
so it is not took in.
a6f22e8356 Show ignored constants as "$N" rather than "?" in pg_stat_statements.
Kyotaro Horiguchi [Thu, 14 Jun 2018 10:15:14 +0000 (19:15 +0900)]
Support PG11
Support PG11. Just adapted to PG11 in this version. No behavior
changes except Parallel hint. PG11 gets many new parallel features and
as a matter of course they made deadly impact on Parallel hint. It no
longer works in the same way with the earlier version except for quite
simple cases.
Kyotaro Horiguchi [Thu, 24 May 2018 09:18:29 +0000 (18:18 +0900)]
Fix a crash bug in case debug_query_string is NULL
pg_hint_plan believed that debug_query_string cannot be null when
parse_analyze is called, but for example in the case under
exec_describe_statement_message, it is not. We see the query string in
pstate even in the case, so use it instead in the case. Since pstate
is storing the query of the lowermost level, we should use
debug_query_string in other cases.
Kyotaro Horiguchi [Mon, 6 Nov 2017 06:28:36 +0000 (15:28 +0900)]
Move expected/ut-W.out to output/ut-W.source
ut-W.out contained an absolute path of my working directory. Moved the
file into output so that we can remove the path replacing with
@abs_srcdir@.
Arthur Zakirov [Thu, 2 Nov 2017 13:10:10 +0000 (16:10 +0300)]
Use @abs_srcdir@ within ut-fdw.source
The file contained absolute path of a specific working directory,
replace them with @abs_srcdir@.
Kyotaro Horiguchi [Mon, 6 Nov 2017 05:38:49 +0000 (14:38 +0900)]
Added and removed some entries in .gitignore
Removed core.c, that is not appropriate.
Added .deps and RPMS.
Kyotaro Horiguchi [Mon, 6 Nov 2017 06:52:53 +0000 (15:52 +0900)]
Remove mark_dummy_rel() from core.c
The function is exposed by the commit
f49842d1ee31b976c681322f76025d7732e860f3 so remove it from the file.
Arthur Zakirov [Thu, 2 Nov 2017 13:00:09 +0000 (16:00 +0300)]
Use pg_strcasecmp instead of strcasecmp for the sake of portability
Back patched to branches PG93-96 and PG10.
Kyotaro Horiguchi [Tue, 10 Oct 2017 02:49:25 +0000 (11:49 +0900)]
Change version to 1.3.0
Kyotaro Horiguchi [Tue, 10 Oct 2017 03:17:58 +0000 (12:17 +0900)]
Added a notice in the manual.
Added an item that explains about letter case handling of object names
in hints.
Kyotaro Horiguchi [Tue, 10 Oct 2017 02:37:39 +0000 (11:37 +0900)]
Prepare for rpm build for PostgreSQL 10
Edited Makefile and SPEC file so that RPM package for PostgreSQL 10 is
built.
Kyotaro Horiguchi [Tue, 10 Oct 2017 02:20:13 +0000 (11:20 +0900)]
Added a regtest for partitioned table.
Added a minimal regression test that checks this works also on
partitioned tables. Currently partitioned tables are handled in almost
the same way with inheritance tables so this would be enough.
Kyotaro Horiguchi [Tue, 10 Oct 2017 01:34:19 +0000 (10:34 +0900)]
Corrected copyright year and a indentation.
Edited a copyright notice in which year is left alone. Correct an
indentation in make_join_rel.c. No functional change.
Kyotaro Horiguchi [Tue, 10 Oct 2017 01:33:00 +0000 (10:33 +0900)]
Reflected changes as of release 10.0.
add_paths_to_append_rel gets modified as of relase 10.0. Reflected it
to core.c
Kyotaro Horiguchi [Wed, 4 Oct 2017 00:46:43 +0000 (09:46 +0900)]
Add a regression about forcing gather-merge.
The previous commit is said to allow enforce gather merge but no
regression for that is not included. This adds that.
Kyotaro Horiguchi [Thu, 27 Jul 2017 10:09:22 +0000 (19:09 +0900)]
Modify expected files.
At 10 beta2, planner seems to get some changes to affect plans used in
regression test. Tentatively modify it just to pass the regression.
Probably it is no problem but should be checked later.
Kyotaro Horiguchi [Thu, 27 Jul 2017 10:07:55 +0000 (19:07 +0900)]
Up to date with 10 beta 2
Make copied code to be up to date with the core as of 10 beta 2.
Kyotaro Horiguchi [Thu, 27 Jul 2017 03:26:55 +0000 (12:26 +0900)]
Support DECLARE CURSOR syntax and added regression for table hinting.
Previous version forgets to treat DECLARE CURSOR. Added that.
Addition to that, regression test for table hinting is added.
Kyotaro Horiguchi [Tue, 6 Jun 2017 10:56:41 +0000 (19:56 +0900)]
Support PostgreSQL 10 beta 1 step 2/2
Changed the logic to tweak parallel execution paths so that it handles
more wider (parallel index scans or gather merge, specifically) range
of parallel paths.
Kyotaro Horiguchi [Tue, 6 Jun 2017 10:23:17 +0000 (19:23 +0900)]
Support PostgreSQL 10 beta 1 step 1/2
The first half of the work to support PostgreSQL 10.
This just makes it can work with the version of PG. So partial index
paths are lost during Parallel hint processing. Regression tests are
heavily modified but most of them just comes from the change of
original planner behavior introduced at the point of Pg10.
Kyotaro Horiguchi [Fri, 19 May 2017 01:11:55 +0000 (10:11 +0900)]
Fix a bug of target query retrieval
The previous patch leaves a bug that get_query_string returns a Query
that JumbleQuery doesn't accept. This fixes that bug.
Kyotaro Horiguchi [Thu, 18 May 2017 09:43:09 +0000 (18:43 +0900)]
Fix a crash bug on complex views when enable_hint_table is on
The Query that planner receives sometimes irrelevant to
debug_query_string. If enable_hint_table is on, pg_hint_plan_planner
normalizes debug_query_string using query-jumble information created
from the irrelevant Query the can lead to crash. To avoid this
situation, retrieve hints in post_parse_analyze_hook, where
corresponding pairs of a query string and a parsed Query.
Kyotaro Horiguchi [Thu, 18 May 2017 01:40:48 +0000 (10:40 +0900)]
Follow the behavioral change of parallel execution
The behavior of parallel planning is changed until 9.6.3, let
expected/ut-W.out to follow that.
Kyotaro Horiguchi [Thu, 18 May 2017 01:28:49 +0000 (10:28 +0900)]
Fix a bug of Parallel hint
During manipulation on pathlist of a GatherPath, foreach was mitakenly
used for a loop using list_delete_cell. This leads to a crash with
SEGV.
Kyotaro Horiguchi [Tue, 21 Feb 2017 00:27:55 +0000 (09:27 +0900)]
Provide upgrade path from 1.1.3
pg_hint_plan96-1.2.0 has an upgrade path from 1.1.3. This adds
requires installation sql file and puts pointers to the files in
Makefile and spec file. It is the only upgrade path of pg_hint_plan
1.2.0 since it works only on PG9.6 and versions earlier than 1.1.3
don't support PG9.6.
Kyotaro Horiguchi [Thu, 16 Feb 2017 10:08:54 +0000 (19:08 +0900)]
Don't apply parallel hint on rels that planner considers unsafe.
pg_hint_plan considered had its own decision logic on whether
parallel-safe or not but it is dangerous and unnecessary. Just follow
planner's decision.
Kyotaro Horiguchi [Thu, 16 Feb 2017 08:41:30 +0000 (17:41 +0900)]
Reconsider hinting process
I rethinked the steps of enforcement considering the parallel planning
code. pg_hint_plan_set_rel_pathlist() gets simpler and easier to read
(maybe).
As the result of this patch, some behaviors has been changed.
- False generation of gahter path caused by hinting on the other side
of joins.
- Scan hints work with Parallel hints. (It's useless for the moment,
though).
Tatsuro Yamada [Thu, 16 Feb 2017 02:00:32 +0000 (11:00 +0900)]
Fix a typo of documentation.
Author: Tatsuro Yamada <yamatattsu@gmail.com>
Kyotaro Horiguchi [Mon, 13 Feb 2017 08:01:49 +0000 (17:01 +0900)]
Don't set_plain_rel_pathlist for other than RTE_RELATION.
pg_hint_plan_set_rel_pathlist can call set_plain_rel_pathlist for
relations other than RTE_RELATION and this results in a crash.
Kyotaro Horiguchi [Mon, 13 Feb 2017 07:56:44 +0000 (16:56 +0900)]
Following a fix in PG9.6.1.
PG9.6.2 gets a fix on planning that affects on the result of
ut-S. ut-S follows that by this fix. As the result ut-S won't success
for 9.6.1 or ealier after this patch.
Kyotaro Horiguchi [Mon, 13 Feb 2017 07:52:45 +0000 (16:52 +0900)]
Fix parallel hint error handling.
Some kind of error of a parallel hint resulted confusing log
output. Addition to that, due to improper assumption of .conf file,
ordinary setup of max_worker_processes caused failure of ut-W. Fixed
them.
Kyotaro Horiguchi [Fri, 10 Feb 2017 08:19:43 +0000 (17:19 +0900)]
Refactor hint application mechamism next step
I removed useless code forgotten to be removed in
add_paths_to_joinrel_wrapper by the previous refactoring. Addition to
that, find_scan_method_hint and find_parallel_hint are found not to
require the third parameter rel, so removed it. Some inadequate tests
are fixed and changed according to change of unclear behaviors.
Kyotaro Horiguchi [Fri, 10 Feb 2017 02:06:05 +0000 (11:06 +0900)]
Improve error handling of Parallel hint
Improve poor error report for Parallel hint.
Kyotaro Horiguchi [Fri, 10 Feb 2017 02:09:52 +0000 (11:09 +0900)]
Fix wrongly lowering max_parallel_workers_per_gather
Number of parallel workers of non-hinted relations is wrongly forced
to be zero even if max_parallel_workers_per_gather is more than
zero. Non-hinted relations get to have max_parallel_workes_per_gather
as set by this fix.
Kyotaro Horiguchi [Thu, 9 Feb 2017 01:50:29 +0000 (10:50 +0900)]
Refactor hint application mechamism
Before we have the new hook set_rel_pathlist_hook, plan enforcment is
performed in two steps, one works only for queries with one relation,
the other works only for joins. Now they are performed once and at
once in set_rel_pathlist_hook. As a byproduct of this refactoring,
index restriction no longer runs for indexes on constraint-excluded
relations and expected/ut-S.out changed a bit.
The major changes by this patch are the followings.
- Remove hook to get_relation_info_hook:
The function for the hook pg_hint_plan_get_relation_info is removed.
- pg_hint_plan_join_search no longer calls rebuild_scan_path. The
function and a copied function set_plain_rel_pathlist are removed.
- delete_indexes has been renamed to restrict_indexes, which
represents the functionality clearer. And the function was
refactored main for readablity.
- Added several regtests for parallel.
Kyotaro Horiguchi [Thu, 9 Feb 2017 00:07:14 +0000 (09:07 +0900)]
Fix default_version of extension control file
default_version in pg_hint_plan.control was left unchanged on updating
version number. Fixed it.
Kyotaro Horiguchi [Thu, 9 Feb 2017 00:06:28 +0000 (09:06 +0900)]
Edit copyright year
Kyotaro Horiguchi [Wed, 8 Feb 2017 01:34:09 +0000 (10:34 +0900)]
Properly change the state of parallel hints.
The state member of parallel hint is forgotten to be chaned on usage.
This fixes it.
Kyotaro Horiguchi [Wed, 8 Feb 2017 01:16:01 +0000 (10:16 +0900)]
Rename some functions and refactor code.
Some functions had a bit confusing naming and inconvenient interfaces.
Renamed such functions and refactored the code.