]> granicus.if.org Git - postgresql/log
postgresql
17 years agoAdd item:
Bruce Momjian [Mon, 2 Apr 2007 20:43:40 +0000 (20:43 +0000)]
Add item:

>
> * Consider reducing memory used for shared buffer reference count
>
>   http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php

17 years agoAdd URL for:
Bruce Momjian [Mon, 2 Apr 2007 20:10:16 +0000 (20:10 +0000)]
Add URL for:

o Allow RETURN to return row or record functions

>  o Allow RETURN to return row or record functions
>    http://archives.postgresql.org/pgsql-patches/2006-08/msg00397.php
>    http://archives.postgresql.org/pgsql-hackers/2006-09/msg00388.php

17 years agoFix check_sql_fn_retval to allow the case where a SQL function declared to
Tom Lane [Mon, 2 Apr 2007 18:49:29 +0000 (18:49 +0000)]
Fix check_sql_fn_retval to allow the case where a SQL function declared to
return void ends with a SELECT, if that SELECT has a single result that is
also of type void.  Without this, it's hard to write a void function that
calls another void function.  Per gripe from Peter.

Back-patch as far as 8.0.

17 years agoDone:
Bruce Momjian [Mon, 2 Apr 2007 17:18:44 +0000 (17:18 +0000)]
Done:

< * Support a data type with specific enumerated values (ENUM)
<
<   http://archives.postgresql.org/pgsql-hackers/2006-08/msg00979.php
<
> * -Support a data type with specific enumerated values (ENUM)

17 years ago> o Add PQexecf() that allows complex parameter substitution
Bruce Momjian [Mon, 2 Apr 2007 17:15:39 +0000 (17:15 +0000)]
>  o Add PQexecf() that allows complex parameter substitution
>
>    http://archives.postgresql.org/pgsql-hackers/2007-03/msg01803.php

17 years agoPut documentation on XML data type and functions in better positions. Add
Peter Eisentraut [Mon, 2 Apr 2007 15:27:02 +0000 (15:27 +0000)]
Put documentation on XML data type and functions in better positions.  Add
some index terms.

17 years agoxcopy can only deal with forward-slashed paths when it's quoted - needed
Magnus Hagander [Mon, 2 Apr 2007 12:11:26 +0000 (12:11 +0000)]
xcopy can only deal with forward-slashed paths when it's quoted - needed
for "vcregress check" to work.
Per report from Dave Page.

17 years agoFix caching of unsuccessful initialization of parser or configuration.
Teodor Sigaev [Mon, 2 Apr 2007 11:42:04 +0000 (11:42 +0000)]
Fix caching of unsuccessful initialization of parser or configuration.
Per report from Listmail <lists@peufeu.com>

17 years agoSupport enum data types. Along the way, use macros for the values of
Tom Lane [Mon, 2 Apr 2007 03:49:42 +0000 (03:49 +0000)]
Support enum data types.  Along the way, use macros for the values of
pg_type.typtype whereever practical.  Tom Dunstan, with some kibitzing
from Tom Lane.

17 years agoUpdate catversion for new XML mapping functions
Peter Eisentraut [Sun, 1 Apr 2007 09:56:02 +0000 (09:56 +0000)]
Update catversion for new XML mapping functions

17 years agoUpdate SQL conformance for SQL to XML mappings
Peter Eisentraut [Sun, 1 Apr 2007 09:42:57 +0000 (09:42 +0000)]
Update SQL conformance for SQL to XML mappings

17 years agoMapping schemas and databases to XML and XML Schema.
Peter Eisentraut [Sun, 1 Apr 2007 09:00:26 +0000 (09:00 +0000)]
Mapping schemas and databases to XML and XML Schema.

Refactor and document the remaining mapping code.

17 years agoComplete cleaning
Peter Eisentraut [Sun, 1 Apr 2007 08:56:58 +0000 (08:56 +0000)]
Complete cleaning

17 years agoAdd some instrumentation to the bgwriter, through the stats collector.
Magnus Hagander [Fri, 30 Mar 2007 18:34:56 +0000 (18:34 +0000)]
Add some instrumentation to the bgwriter, through the stats collector.
New view pg_stat_bgwriter, and the functions required to build it.

17 years agoUpdate Japanese FAQ.
Bruce Momjian [Fri, 30 Mar 2007 03:46:53 +0000 (03:46 +0000)]
Update Japanese FAQ.

Jun Kuwamura

17 years agoUpdate SSL description for when SSL root.crt/server.crt is required;
Bruce Momjian [Fri, 30 Mar 2007 03:19:02 +0000 (03:19 +0000)]
Update SSL description for when SSL root.crt/server.crt is required;
add link to libpq SSL does from server docs.

Backpatch to 8.2.X.

17 years agoFix oversight in coding of _bt_start_vacuum: we can't assume that the LWLock
Tom Lane [Fri, 30 Mar 2007 00:12:59 +0000 (00:12 +0000)]
Fix oversight in coding of _bt_start_vacuum: we can't assume that the LWLock
will be released by transaction abort before _bt_end_vacuum gets called.
If either of these "can't happen" errors actually happened, we'd freeze up
trying to acquire an already-held lock.  Latest word is that this does
not explain Martin Pitt's trouble report, but it still looks like a bug.

17 years agopg_char_to_encoding() was redundant in initdb because
Bruce Momjian [Thu, 29 Mar 2007 22:46:42 +0000 (22:46 +0000)]
pg_char_to_encoding() was redundant in initdb because
pg_valid_server_encoding() returns the same result if the encoding is
valid.

ITAGAKI Takahiro

17 years agoAdd to:
Bruce Momjian [Thu, 29 Mar 2007 21:44:39 +0000 (21:44 +0000)]
Add to:

* Reduce storage space for small NUMERICs
>   http://archives.postgresql.org/pgsql-patches/2007-02/msg00505.php

17 years agoAdd:
Bruce Momjian [Thu, 29 Mar 2007 21:43:34 +0000 (21:43 +0000)]
Add:

> * Reduce storage space for small NUMERICs
>
>   http://archives.postgresql.org/pgsql-hackers/2007-02/msg01331.php
>

17 years agoInstall include files
Magnus Hagander [Thu, 29 Mar 2007 20:48:26 +0000 (20:48 +0000)]
Install include files

17 years agoUpdate:
Bruce Momjian [Thu, 29 Mar 2007 20:24:34 +0000 (20:24 +0000)]
Update:

<  o Add more logical syntax CLUSTER table USING index;
>  o Add more logical syntax CLUSTER table USING index;

17 years agoUpdate item:
Bruce Momjian [Thu, 29 Mar 2007 20:23:56 +0000 (20:23 +0000)]
Update item:

        o Add more logical syntax CLUSTER table USING index;
<  o Add more logical syntax CLUSTER table ORDER BY index;
>  o Add more logical syntax CLUSTER table USING index;

17 years agoexec_parse_message neglected to copy parameter type array into the
Tom Lane [Thu, 29 Mar 2007 19:10:10 +0000 (19:10 +0000)]
exec_parse_message neglected to copy parameter type array into the
required memory context when handling client-specified parameter types
for an unnamed statement.  Per report from Kris Jurka.

17 years agoMake ECPG regression tests use native threading instead of pthreads, now that
Magnus Hagander [Thu, 29 Mar 2007 15:30:52 +0000 (15:30 +0000)]
Make ECPG regression tests use native threading instead of pthreads, now that
 ecpglib supports it.
Change configure (patch from Bruce) and msvc build system to no longer require
 pthreads on win32, since all parts of postgresql can be thread-safe using the
 native platform functions.

17 years ago- Added patch by Magnus Hagander <magnus@hagander.net> to use native
Michael Meskes [Thu, 29 Mar 2007 12:02:24 +0000 (12:02 +0000)]
- Added patch by Magnus Hagander <magnus@hagander.net> to use native
  win32 threads.
- Fixed regression tests to run threading tests.

17 years agoTeach CLUSTER to skip writing WAL if not needed (ie, not using archiving)
Tom Lane [Thu, 29 Mar 2007 00:15:39 +0000 (00:15 +0000)]
Teach CLUSTER to skip writing WAL if not needed (ie, not using archiving)
--- Simon.
Also, code review and cleanup for the previous COPY-no-WAL patches --- Tom.

17 years agoCode cleanup for the new regexp UDFs: we can hardcode the OID and some
Neil Conway [Wed, 28 Mar 2007 22:59:37 +0000 (22:59 +0000)]
Code cleanup for the new regexp UDFs: we can hardcode the OID and some
properties of the "text" type, and then simplify the code accordingly.
Patch from Jeremy Drake.

17 years agoReplace 4-clause licensed blf.[ch] with blowfish implementation
Neil Conway [Wed, 28 Mar 2007 22:48:58 +0000 (22:48 +0000)]
Replace 4-clause licensed blf.[ch] with blowfish implementation
from PuTTY with is under minimal BSD/MIT license. Marko Kreen.

17 years agoAdd the "recheck" logic to autovacuum worker code. The worker first builds
Alvaro Herrera [Wed, 28 Mar 2007 22:17:12 +0000 (22:17 +0000)]
Add the "recheck" logic to autovacuum worker code.  The worker first builds
its table list and then rechecks pgstat before vacuuming each table to
verify that no one has vacuumed the table in the meantime.

In the current autovacuum world this only means that a worker will not
vacuum a table that a user has vacuumed manually after the worker started.
When support for multiple autovacuum workers is introduced, this will reduce
the probability of simultaneous workers on the same database doing redundant
work.

17 years agoFix compiler warnings in ereport messages on mingw.
Magnus Hagander [Wed, 28 Mar 2007 08:06:11 +0000 (08:06 +0000)]
Fix compiler warnings in ereport messages on mingw.
ITAGAKI Takahiro

17 years agoAdd URL for:
Bruce Momjian [Wed, 28 Mar 2007 02:17:45 +0000 (02:17 +0000)]
Add URL for:

* Automatically create rules on views so they are updateable, per SQL99
>   http://archives.postgresql.org/pgsql-patches/2006-08/msg00255.php

17 years agoFix uninitialized-variable bug.
Tom Lane [Wed, 28 Mar 2007 01:28:34 +0000 (01:28 +0000)]
Fix uninitialized-variable bug.

17 years agoFix array coercion expressions to ensure that the correct volatility is
Tom Lane [Tue, 27 Mar 2007 23:21:12 +0000 (23:21 +0000)]
Fix array coercion expressions to ensure that the correct volatility is
seen by code inspecting the expression.  The best way to do this seems
to be to drop the original representation as a function invocation, and
instead make a special expression node type that represents applying
the element-type coercion function to each array element.  In this way
the element function is exposed and will be checked for volatility.
Per report from Guillaume Smet.

17 years agoInstall import libraries used to link to libpq, ecpg and the backend.
Magnus Hagander [Tue, 27 Mar 2007 21:47:10 +0000 (21:47 +0000)]
Install import libraries used to link to libpq, ecpg and the backend.

17 years agoCosmetic changes: rename some struct fields, and move the fetching of pgstat
Alvaro Herrera [Tue, 27 Mar 2007 20:36:03 +0000 (20:36 +0000)]
Cosmetic changes: rename some struct fields, and move the fetching of pgstat
table entries to a separate routine.  Don't pass the pgstat database entry to
do_autovacuum; rather, have it fetch it by itself.

17 years agoRemove item due to lack of interest:
Bruce Momjian [Tue, 27 Mar 2007 17:27:01 +0000 (17:27 +0000)]
Remove item due to lack of interest:

< * %Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
<   pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
<
<   These would be for application use, not for use by pg_dump.
<

17 years agoAdd:
Bruce Momjian [Tue, 27 Mar 2007 15:36:40 +0000 (15:36 +0000)]
Add:

>
> * Allow configuration of backend priorities via the operating system
>
>   Though backend priorities make priority inversion during lock
>   waits possible, research shows that this is not a huge problem.
>   http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php

17 years agoFix typo in Makefile.
Bruce Momjian [Tue, 27 Mar 2007 14:29:51 +0000 (14:29 +0000)]
Fix typo in Makefile.

Marko Kreen

17 years agoFix typo, ensable -> enable, per Steve Gieseking.
Tom Lane [Tue, 27 Mar 2007 03:25:28 +0000 (03:25 +0000)]
Fix typo, ensable -> enable, per Steve Gieseking.

17 years agoAdd documentation for pulling the CVS repository using rsync, and
Bruce Momjian [Tue, 27 Mar 2007 01:45:22 +0000 (01:45 +0000)]
Add documentation for pulling the CVS repository using rsync, and
mention cvsup last.

17 years agoUpdate:
Bruce Momjian [Mon, 26 Mar 2007 23:48:43 +0000 (23:48 +0000)]
Update:

* Improve dead row detection during multi-statement transactions usage

17 years agoAdd:
Bruce Momjian [Mon, 26 Mar 2007 23:47:53 +0000 (23:47 +0000)]
Add:

> * Increase the number of rows recognized as dead during multi-statement
>   transactions
>
>   http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
>

17 years agoRemove Andrew Yu copyright, with permission from author.
Bruce Momjian [Mon, 26 Mar 2007 21:51:15 +0000 (21:51 +0000)]
Remove Andrew Yu copyright, with permission from author.

17 years agoRemove advertising clause from Berkeley BSD-licensed files, per
Bruce Momjian [Mon, 26 Mar 2007 21:44:11 +0000 (21:44 +0000)]
Remove advertising clause from Berkeley BSD-licensed files, per
instructions from Berkeley.

17 years agoProperly fix PORTNAME in configure.
Bruce Momjian [Mon, 26 Mar 2007 21:30:56 +0000 (21:30 +0000)]
Properly fix PORTNAME in configure.

17 years agoUser PORTNAME for win32 tests, rather than $template, for consistency.
Bruce Momjian [Mon, 26 Mar 2007 19:46:56 +0000 (19:46 +0000)]
User PORTNAME for win32 tests, rather than $template, for consistency.

17 years agoFix seriously broken markup for libpq-envars cross-references.
Tom Lane [Mon, 26 Mar 2007 17:23:37 +0000 (17:23 +0000)]
Fix seriously broken markup for libpq-envars cross-references.

17 years agoAllow non-superuser database owners to create procedural languages.
Tom Lane [Mon, 26 Mar 2007 16:58:41 +0000 (16:58 +0000)]
Allow non-superuser database owners to create procedural languages.
A DBA is allowed to create a language in his database if it's marked
"tmpldbacreate" in pg_pltemplate.  The factory default is that this is set
for all standard trusted languages, but of course a superuser may adjust
the settings.  In service of this, add the long-foreseen owner column to
pg_language; renaming, dropping, and altering owner of a PL now follow
normal ownership rules instead of being superuser-only.
Jeremy Drake, with some editorialization by Tom Lane.

17 years agoAdd checking of end of line in parsing stopword list. Thanks to sharp eyes of Tom...
Teodor Sigaev [Mon, 26 Mar 2007 13:57:07 +0000 (13:57 +0000)]
Add checking of end of line in parsing stopword list. Thanks to sharp eyes of Tom lane

17 years agoFix stopword and synonym files parsing bug in MSVC build, per report from
Teodor Sigaev [Mon, 26 Mar 2007 12:25:35 +0000 (12:25 +0000)]
Fix stopword and synonym files parsing bug in MSVC build, per report from
Magnus Hagander. Also, now it ignores space symbol after stopwords.

17 years agoFix pg_wchar_table's maxmblen field of EUC_CN, EUC_TW, MULE_INTERNAL
Tatsuo Ishii [Mon, 26 Mar 2007 11:15:13 +0000 (11:15 +0000)]
Fix pg_wchar_table's maxmblen field of EUC_CN, EUC_TW, MULE_INTERNAL
and GB18030. patches from ITAGAKI Takahiro.

17 years agoSeems some people have been forgetting to run autoheader.
Tom Lane [Mon, 26 Mar 2007 02:38:22 +0000 (02:38 +0000)]
Seems some people have been forgetting to run autoheader.

17 years agoAdd -lcrypto as one of the possible link dependencies of libkrb5.
Tom Lane [Mon, 26 Mar 2007 02:37:17 +0000 (02:37 +0000)]
Add -lcrypto as one of the possible link dependencies of libkrb5.
Per report from Jim Rosenberg.  This possibly should get back-patched,
but I'm a bit suspicious of it still because of the lack of prior reports.

17 years agoRemove assertion that constraint_exclusion risks wrong answers if
Tom Lane [Mon, 26 Mar 2007 01:41:57 +0000 (01:41 +0000)]
Remove assertion that constraint_exclusion risks wrong answers if
table constraints are changed; this is no longer true now that we
have a plan invalidation mechanism.

17 years agoFix plancache's invalidation callback to do the right thing for a SI
Tom Lane [Mon, 26 Mar 2007 00:36:19 +0000 (00:36 +0000)]
Fix plancache's invalidation callback to do the right thing for a SI
reset event, namely invalidate everything.  This oversight probably
explains the rare failures that some buildfarm machines have been
showing for the plancache regression test.

17 years agoMake _SPI_execute_plan pass the query source string down to ProcessUtility
Tom Lane [Sun, 25 Mar 2007 23:42:43 +0000 (23:42 +0000)]
Make _SPI_execute_plan pass the query source string down to ProcessUtility
if possible.  I had left this undone in the first pass at the API change
for ProcessUtility, but forgot to revisit it after the plancache changes
made it possible to do it.

17 years agoRemove the prohibition on executing cursor commands through SPI_execute.
Tom Lane [Sun, 25 Mar 2007 23:27:59 +0000 (23:27 +0000)]
Remove the prohibition on executing cursor commands through SPI_execute.
Vadim had included this restriction in the original design of the SPI code,
but I'm darned if I can see a reason for it.

I left the macro definition of SPI_ERROR_CURSOR in place, so as not to
needlessly break any SPI callers that are checking for it, but that code
will never actually be returned anymore.

17 years agoClean up the representation of special snapshots by including a "method
Tom Lane [Sun, 25 Mar 2007 19:45:14 +0000 (19:45 +0000)]
Clean up the representation of special snapshots by including a "method
pointer" in every Snapshot struct.  This allows removal of the case-by-case
tests in HeapTupleSatisfiesVisibility, which should make it a bit faster
(I didn't try any performance tests though).  More importantly, we are no
longer violating portable C practices by assuming that small integers are
distinct from all pointer values, and HeapTupleSatisfiesDirty no longer
has a non-reentrant API involving side-effects on a global variable.

There were a couple of places calling HeapTupleSatisfiesXXX routines
directly rather than through the HeapTupleSatisfiesVisibility macro.
Since these places had to be changed anyway, I chose to make them go
through the macro for uniformity.

Along the way I renamed HeapTupleSatisfiesSnapshot to HeapTupleSatisfiesMVCC
to emphasize that it's only used with MVCC-type snapshots.  I was sorely
tempted to rename HeapTupleSatisfiesVisibility to HeapTupleSatisfiesSnapshot,
but forebore for the moment to avoid confusion and reduce the likelihood that
this patch breaks some of the pending patches.  Might want to reconsider
doing that later.

17 years agoAdd new encoding EUC_JIS_2004 and SHIFT_JIS_2004,
Tatsuo Ishii [Sun, 25 Mar 2007 11:56:04 +0000 (11:56 +0000)]
Add new encoding EUC_JIS_2004 and SHIFT_JIS_2004,
along with new conversions among EUC_JIS_2004, SHIFT_JIS_2004 and UTF-8.
catalog version has been bump up.

17 years agoAdd:
Bruce Momjian [Sun, 25 Mar 2007 02:21:19 +0000 (02:21 +0000)]
Add:

>
> * Allow BEFORE INSERT triggers on views
>
>   http://archives.postgresql.org/pgsql-general/2007-02/msg01466.php

17 years agoAdd:
Bruce Momjian [Sat, 24 Mar 2007 23:12:34 +0000 (23:12 +0000)]
Add:

>  o Add more logical syntax CLUSTER table ORDER BY index;
>    support current syntax for backward compatibility

17 years agoSupport for installing NLS files, and update support to use gettext
Magnus Hagander [Sat, 24 Mar 2007 22:16:49 +0000 (22:16 +0000)]
Support for installing NLS files, and update support to use gettext
from gnuwin32.

17 years agoDocument that LDAP URLs should be double-quoted in pg_hba.conf because
Bruce Momjian [Sat, 24 Mar 2007 21:46:23 +0000 (21:46 +0000)]
Document that LDAP URLs should be double-quoted in pg_hba.conf because
commas are often present in the URL.

Backpatch to 8.2.X.

17 years agoInstall contrib sql and readme files
Magnus Hagander [Sat, 24 Mar 2007 15:28:48 +0000 (15:28 +0000)]
Install contrib sql and readme files

17 years agoProperly parse the name of contrib modules that aren't named the same
Magnus Hagander [Sat, 24 Mar 2007 14:13:27 +0000 (14:13 +0000)]
Properly parse the name of contrib modules that aren't named the same
way as their directory (notably xml2/pgxml and intarray/_int)

17 years agoSet the node properly, per Tom.
Alvaro Herrera [Fri, 23 Mar 2007 21:57:10 +0000 (21:57 +0000)]
Set the node properly, per Tom.

17 years agoSeparate the code to start a new worker into its own function. The code is
Alvaro Herrera [Fri, 23 Mar 2007 21:45:17 +0000 (21:45 +0000)]
Separate the code to start a new worker into its own function.  The code is
exactly the same, modulo whitespace.

17 years agoSeparate fetch of pg_autovacuum tuple into its own function.
Alvaro Herrera [Fri, 23 Mar 2007 21:23:13 +0000 (21:23 +0000)]
Separate fetch of pg_autovacuum tuple into its own function.

17 years agoWe no longer need to palloc the VacuumStmt node; keeping it on the stack is
Alvaro Herrera [Fri, 23 Mar 2007 20:56:39 +0000 (20:56 +0000)]
We no longer need to palloc the VacuumStmt node; keeping it on the stack is
simpler.

17 years agoAdjust DatumGetBool macro so that it isn't fooled by garbage in the Datum
Tom Lane [Fri, 23 Mar 2007 20:24:41 +0000 (20:24 +0000)]
Adjust DatumGetBool macro so that it isn't fooled by garbage in the Datum
to the left of the actual bool value.  While in most cases there won't be
any, our support for old-style user-defined functions violates the C spec
to the extent of calling functions that might return char or short through
a function pointer declared to return "char *", which we then coerce to
Datum.  It is not surprising that the result might contain garbage
high-order bits ... what is surprising is that we didn't see such cases
long ago.  Per report from Magnus.

17 years agoFix plancache so that any required replanning is done with the same
Tom Lane [Fri, 23 Mar 2007 19:53:52 +0000 (19:53 +0000)]
Fix plancache so that any required replanning is done with the same
search_path that was active when the plan was first made.  To do this,
improve namespace.c to support a stack of "override" search path settings
(we must have a stack since nested replan events are entirely possible).
This facility replaces the "special namespace" hack formerly used by
CREATE SCHEMA, and should be able to support per-function search path
settings as well.

17 years agoAllow 4 bytes UTF-8 (UCS-4 range 00010000-001FFFFF)
Tatsuo Ishii [Fri, 23 Mar 2007 13:51:30 +0000 (13:51 +0000)]
Allow 4 bytes UTF-8 (UCS-4 range 00010000-001FFFFF)
This is necessary to support JIS X 0213 <--> UTF-8 conversion.

17 years agoForgot commit: support for special-cases in pgcrypto
Magnus Hagander [Fri, 23 Mar 2007 10:05:34 +0000 (10:05 +0000)]
Forgot commit: support for special-cases in pgcrypto

17 years agoSupport for running contribcheck on msvc
Magnus Hagander [Fri, 23 Mar 2007 09:57:55 +0000 (09:57 +0000)]
Support for running contribcheck on msvc

17 years agoGenerate SQL files for /contrib (based on .sql.in)
Magnus Hagander [Fri, 23 Mar 2007 09:53:33 +0000 (09:53 +0000)]
Generate SQL files for /contrib (based on .sql.in)

17 years agoAdd define to exclude configured libraries, to be able to easily build a
Magnus Hagander [Fri, 23 Mar 2007 08:43:51 +0000 (08:43 +0000)]
Add define to exclude configured libraries, to be able to easily build a
stripped down version of libpq. To be used by the installer.

17 years agoRemove headers for old sysv shmem emulation that I forgot.
Magnus Hagander [Fri, 23 Mar 2007 08:30:55 +0000 (08:30 +0000)]
Remove headers for old sysv shmem emulation that I forgot.
Also remove headers for old sysv semaphore emulation that were forgotten
when that was changed about a year ago.

17 years agoCleanup for procarray.c.
Bruce Momjian [Fri, 23 Mar 2007 03:16:39 +0000 (03:16 +0000)]
Cleanup for procarray.c.

17 years agoAdd:
Bruce Momjian [Thu, 22 Mar 2007 20:55:19 +0000 (20:55 +0000)]
Add:

>
> * Fix cases where invalid byte encodings are accepted by the database,
>   but throw an error on SELECT
>
>   http://archives.postgresql.org/pgsql-hackers/2007-03/msg00767.php

17 years agoAdd comment that pg_dump 'append' format is used only by pg_dump, per
Bruce Momjian [Thu, 22 Mar 2007 20:47:12 +0000 (20:47 +0000)]
Add comment that pg_dump 'append' format is used only by pg_dump, per
Dave Page.

17 years agoAdd to TODO:
Bruce Momjian [Thu, 22 Mar 2007 20:38:42 +0000 (20:38 +0000)]
Add to TODO:

CLUSTER:
>
>  o %Add VERBOSE option to report tables as they are processed,
>    like VACUUM VERBOSE

17 years agoCommit newest version of xmlpath().
Bruce Momjian [Thu, 22 Mar 2007 20:26:30 +0000 (20:26 +0000)]
Commit newest version of xmlpath().

Nikolay Samokhvalov

17 years agoIn pg_dump, strcasecmp -> pg_strcasecmp.
Bruce Momjian [Thu, 22 Mar 2007 20:18:49 +0000 (20:18 +0000)]
In pg_dump, strcasecmp -> pg_strcasecmp.

17 years agoAdd xmlpath() to evaluate XPath expressions, with namespaces support.
Bruce Momjian [Thu, 22 Mar 2007 20:14:58 +0000 (20:14 +0000)]
Add xmlpath() to evaluate XPath expressions, with namespaces support.

Nikolay Samokhvalov

17 years agoArrange for PreventTransactionChain to reject commands submitted as part
Tom Lane [Thu, 22 Mar 2007 19:55:04 +0000 (19:55 +0000)]
Arrange for PreventTransactionChain to reject commands submitted as part
of a multi-statement simple-Query message.  This bug goes all the way
back, but unfortunately is not nearly so easy to fix in existing releases;
it is only the recent ProcessUtility API change that makes it fixable in
HEAD.  Per report from William Garrison.

17 years agoAllow the pgstat process to restart immediately after a receiving
Bruce Momjian [Thu, 22 Mar 2007 19:53:31 +0000 (19:53 +0000)]
Allow the pgstat process to restart immediately after a receiving
SIGQUIT signal, rather than waiting for PGSTAT_RESTART_INTERVAL.

17 years agoAllow DROP TABLESPACE to succeed (with a warning) if the pg_tblspc symlink
Tom Lane [Thu, 22 Mar 2007 19:51:44 +0000 (19:51 +0000)]
Allow DROP TABLESPACE to succeed (with a warning) if the pg_tblspc symlink
doesn't exist.  This allows DROP to be used to clean out the pg_tablespace
catalog entry in a situation where a previous DROP attempt failed before
committing but after having removed the directories and symlink.

Per report from William Garrison.  Even though his test case depends on an
unrelated bug in PreventTransactionChain, it's certainly possible for this
situation to arise due to other problems, eg a system crash at just the
right time.

17 years agoProperly enforce pg_dump -F formation options; only single letter or
Bruce Momjian [Thu, 22 Mar 2007 19:42:02 +0000 (19:42 +0000)]
Properly enforce pg_dump -F formation options; only single letter or
full words support, per report from Mark Stosberg.

17 years agoRemove the currently unused FRONTEND case in dllist.c. This allows the usage
Alvaro Herrera [Thu, 22 Mar 2007 18:57:52 +0000 (18:57 +0000)]
Remove the currently unused FRONTEND case in dllist.c.  This allows the usage
of palloc instead of malloc, which means a list can be freed simply by deleting
the memory context that contains it.

17 years agoFix parser bug on Windows with UTF8 encoding and C locale, the reason was
Teodor Sigaev [Thu, 22 Mar 2007 15:58:24 +0000 (15:58 +0000)]
Fix parser bug on Windows with UTF8 encoding and C locale, the reason was
sizeof(wchar_t) = 2 instead of 4.

17 years agoRemove tabs in SGML files.
Bruce Momjian [Thu, 22 Mar 2007 15:46:56 +0000 (15:46 +0000)]
Remove tabs in SGML files.

17 years agoRemove tabs from SGML file.
Bruce Momjian [Thu, 22 Mar 2007 15:45:56 +0000 (15:45 +0000)]
Remove tabs from SGML file.

17 years agoForgot to add file needed for PL regression tests
Magnus Hagander [Thu, 22 Mar 2007 13:43:02 +0000 (13:43 +0000)]
Forgot to add file needed for PL regression tests

17 years agoFix some problems with selectivity estimation for partial indexes.
Tom Lane [Wed, 21 Mar 2007 22:18:12 +0000 (22:18 +0000)]
Fix some problems with selectivity estimation for partial indexes.

First, genericcostestimate() was being way too liberal about including
partial-index conditions in its selectivity estimate, resulting in
substantial underestimates for situations such as an indexqual "x = 42"
used with an index on x "WHERE x >= 40 AND x < 50".  While the code is
intentionally set up to favor selecting partial indexes when available,
this was too much...

Second, choose_bitmap_and() was likewise easily fooled by cases of this
type, since it would similarly think that the partial index had selectivity
independent of the indexqual.

Fixed by using predicate_implied_by() rather than simple equality checks
to determine redundancy.  This is a good deal more expensive but I don't
see much alternative.  At least the extra cost is only paid when there's
actually a partial index under consideration.

Per report from Jeff Davis.  I'm not going to risk back-patching this,
though.

17 years agoAdd:
Bruce Momjian [Wed, 21 Mar 2007 21:25:46 +0000 (21:25 +0000)]
Add:

<
<
>  o During index creation, pre-sort the tuples to improve build speed
>
>    http://archives.postgresql.org/pgsql-hackers/2007-03/msg01199.php
>

17 years agoRemove TODO item, not wanted:
Bruce Momjian [Wed, 21 Mar 2007 21:21:21 +0000 (21:21 +0000)]
Remove TODO item, not wanted:

< * Add NUMERIC division operator that doesn't round?
<
<   Currently NUMERIC _rounds_ the result to the specified precision.
<   This means division can return a result that multiplied by the
<   divisor is greater than the dividend, e.g. this returns a value > 10:
<
<     SELECT (10::numeric(2,0) / 6::numeric(2,0))::numeric(2,0) * 6;
<
<   The positive modulus result returned by NUMERICs might be considered
<   inaccurate, in one sense.
<

17 years agoAdd documentation about vcregress.
Magnus Hagander [Wed, 21 Mar 2007 19:22:52 +0000 (19:22 +0000)]
Add documentation about vcregress.

17 years agoAdd URL for:
Bruce Momjian [Wed, 21 Mar 2007 16:43:31 +0000 (16:43 +0000)]
Add URL for:

* Add locale-aware MONEY type, and support multiple currencies
  http://archives.postgresql.org/pgsql-hackers/2007-03/msg01181.php

17 years agoAdd support for running regression tests on procedural languages
Magnus Hagander [Wed, 21 Mar 2007 16:21:40 +0000 (16:21 +0000)]
Add support for running regression tests on procedural languages