]> granicus.if.org Git - postgresql/log
postgresql
17 years agoMake autovacuum behavior more agressive, per discussion on hackers list
Bruce Momjian [Sat, 2 Sep 2006 23:12:16 +0000 (23:12 +0000)]
Make autovacuum behavior more agressive, per discussion on hackers list
--- was part of autovacuum default 'on' patch that was reverted, but we
want this part.

Peter Eisentraut

17 years agoUpdate postgresql.conf line for default superuser_reserved_connections.
Bruce Momjian [Sat, 2 Sep 2006 23:08:36 +0000 (23:08 +0000)]
Update postgresql.conf line for default superuser_reserved_connections.

17 years agoChange "superuser_reserved_connections" default to 3, because of
Bruce Momjian [Sat, 2 Sep 2006 23:04:20 +0000 (23:04 +0000)]
Change "superuser_reserved_connections" default to 3, because of
possible autovacuum use.

17 years agoAdd URL for autovacuum default:
Bruce Momjian [Sat, 2 Sep 2006 22:59:00 +0000 (22:59 +0000)]
Add URL for autovacuum default:

>
>    http://archives.postgresql.org/pgsql-hackers/2006-08/msg01852.php

17 years agoUse '' rather than \' for literal single quotes in strings in
Bruce Momjian [Sat, 2 Sep 2006 22:03:30 +0000 (22:03 +0000)]
Use '' rather than \' for literal single quotes in strings in
/contrib/tsearch2.

Teodor Sigaev

17 years agoWording improvements for archive_timeout.
Bruce Momjian [Sat, 2 Sep 2006 21:11:26 +0000 (21:11 +0000)]
Wording improvements for archive_timeout.

17 years agoAdded async query capability. Original patch by
Joe Conway [Sat, 2 Sep 2006 21:11:15 +0000 (21:11 +0000)]
Added async query capability. Original patch by
Kai Londenberg, modified by Joe Conway

17 years agoSmall code cleanup for recent UPDATE SET (values) patch.
Bruce Momjian [Sat, 2 Sep 2006 20:52:01 +0000 (20:52 +0000)]
Small code cleanup for recent UPDATE SET (values) patch.

17 years agoAdd UPDATE tab SET ROW (col, ...) = (val, ...) for updating
Bruce Momjian [Sat, 2 Sep 2006 20:34:47 +0000 (20:34 +0000)]
Add UPDATE tab SET ROW (col, ...) = (val, ...) for updating
multiple columns

Susanne Ebrecht

17 years agoUpdate:
Bruce Momjian [Sat, 2 Sep 2006 20:31:50 +0000 (20:31 +0000)]
Update:

< Last updated: Sat Sep  2 08:31:04 EDT 2006
> Last updated: Sat Sep  2 16:31:46 EDT 2006
<  o Allow UPDATE tab SET ROW (col, ...) = (...) for updating multiple
<    columns
>  o -Allow UPDATE tab SET ROW (col, ...) = (val, ...) for updating
>    multiple columns
>  o Allow UPDATE tab SET ROW (col, ...) = (SELECT...)
<           A subselect can also be used as the value source.

17 years agoChange FETCH/MOVE to use int8.
Bruce Momjian [Sat, 2 Sep 2006 18:17:18 +0000 (18:17 +0000)]
Change FETCH/MOVE to use int8.

Dhanaraj M

17 years agoRevert as not needed/inconsistent with SQL REINDEX:
Bruce Momjian [Sat, 2 Sep 2006 17:10:17 +0000 (17:10 +0000)]
Revert as not needed/inconsistent with SQL REINDEX:

  Suppress some NOTICE messages from REINDEX command.

  Euler Taveira de Oliveira

17 years agoRemove GUC_REPORT for new "server_version_num" GUC variable. Added
Bruce Momjian [Sat, 2 Sep 2006 17:08:10 +0000 (17:08 +0000)]
Remove GUC_REPORT for new "server_version_num" GUC variable.  Added
overhead for every connection, per Tom.

17 years agoApply a simple solution to the problem of making INSERT/UPDATE/DELETE
Tom Lane [Sat, 2 Sep 2006 17:06:52 +0000 (17:06 +0000)]
Apply a simple solution to the problem of making INSERT/UPDATE/DELETE
RETURNING play nice with views/rules.  To wit, have the rule rewriter
rewrite any RETURNING clause found in a rule to produce what the rule's
triggering query asked for in its RETURNING clause, in particular drop
the RETURNING clause if no RETURNING in the triggering query.  This
leaves the responsibility for knowing how to produce the view's output
columns on the rule author, without requiring any fundamental changes
in rule semantics such as adding new rule event types would do.  The
initial implementation constrains things to ensure that there is
exactly one, unconditionally invoked RETURNING clause among the rules
for an event --- later we might be able to relax that, but for a post
feature freeze fix it seems better to minimize how much invention we do.
Per gripe from Jaime Casanova.

17 years agoAdd functions to /contrib/pgstattuple that show index statistics and
Bruce Momjian [Sat, 2 Sep 2006 17:05:29 +0000 (17:05 +0000)]
Add functions to /contrib/pgstattuple that show index statistics and
index page contents.

Satoshi Nagayasu

17 years agoAdd new variable "server_version_num", which is almost the same as
Bruce Momjian [Sat, 2 Sep 2006 13:12:50 +0000 (13:12 +0000)]
Add new variable "server_version_num", which is almost the same as
"server_version" but uses the handy PG_VERSION_NUM which allows apps to
do things like if ($version >= 80200) without having to parse apart the
value of server_version themselves.

Greg Sabino Mullane greg@turnstep.com

17 years agoUpdate wording:
Bruce Momjian [Sat, 2 Sep 2006 12:31:10 +0000 (12:31 +0000)]
Update wording:

<  o -Allow PL/python to composite types and result sets
>  o -Allow PL/python to return composite types and result sets

17 years agoAllow PL/python to return composite types and result sets
Bruce Momjian [Sat, 2 Sep 2006 12:30:01 +0000 (12:30 +0000)]
Allow PL/python to return composite types and result sets

Sven Suursoho

17 years agoDone:
Bruce Momjian [Sat, 2 Sep 2006 12:29:18 +0000 (12:29 +0000)]
Done:

                o -Allow PL/python to composite types and result sets

17 years agoSuppress some NOTICE messages from REINDEX command.
Bruce Momjian [Sat, 2 Sep 2006 02:43:07 +0000 (02:43 +0000)]
Suppress some NOTICE messages from REINDEX command.

Euler Taveira de Oliveira

17 years agoClean up rather sloppy fix in HEAD for the ancient bug that CREATE CONVERSION
Tom Lane [Thu, 31 Aug 2006 17:31:33 +0000 (17:31 +0000)]
Clean up rather sloppy fix in HEAD for the ancient bug that CREATE CONVERSION
didn't create a dependency from the new conversion to its schema.  Back-patch
to all supported releases.

17 years agoRepair interpretation of GB as MB.
Peter Eisentraut [Thu, 31 Aug 2006 15:10:51 +0000 (15:10 +0000)]
Repair interpretation of GB as MB.

17 years agoAttibution addition: Add Karel Zak also for COPY SELECT.
Bruce Momjian [Thu, 31 Aug 2006 03:17:50 +0000 (03:17 +0000)]
Attibution addition:  Add Karel Zak also for COPY SELECT.

17 years agoCorrect attibution:
Bruce Momjian [Thu, 31 Aug 2006 00:35:32 +0000 (00:35 +0000)]
Correct attibution:

COPY SELECT work done by Zoltan Boszormenyi

17 years agoDone:
Bruce Momjian [Thu, 31 Aug 2006 00:34:34 +0000 (00:34 +0000)]
Done:

< Last updated: Tue Aug 29 12:21:52 EDT 2006
> Last updated: Wed Aug 30 20:34:28 EDT 2006
<  o Allow COPY (SELECT ...) TO 'filename'
<
<    COPY should also be able to output views using COPY (SELECT
<    * FROM view) TO 'filename' internally.
<    http://archives.postgresql.org/pgsql-patches/2005-09/msg00148.php
>  o -Allow COPY (SELECT ...) TO 'filename'

17 years agoExtend COPY to support COPY (SELECT ...) TO ...
Tom Lane [Wed, 30 Aug 2006 23:34:22 +0000 (23:34 +0000)]
Extend COPY to support COPY (SELECT ...) TO ...

Bernd Helmle

17 years agoUpdate logging of prepare/execute syntax, per comments from Guillaume Smet.
Bruce Momjian [Wed, 30 Aug 2006 18:22:02 +0000 (18:22 +0000)]
Update logging of prepare/execute syntax, per comments from Guillaume Smet.

17 years agoFix things so that fopen's, not only open's, pass FILE_SHARE_DELETE
Tom Lane [Wed, 30 Aug 2006 18:06:27 +0000 (18:06 +0000)]
Fix things so that fopen's, not only open's, pass FILE_SHARE_DELETE
and other special flags on Windows.  May fix intermittent 'Permission
denied' errors.  Magnus Hagander

17 years agoDept. of second thoughts: if query fails part way through, shut down
Tom Lane [Tue, 29 Aug 2006 22:48:55 +0000 (22:48 +0000)]
Dept. of second thoughts: if query fails part way through, shut down
the pager before reporting the error.

17 years agoCreate a FETCH_COUNT parameter that causes psql to execute SELECT-like
Tom Lane [Tue, 29 Aug 2006 22:25:08 +0000 (22:25 +0000)]
Create a FETCH_COUNT parameter that causes psql to execute SELECT-like
queries via a cursor, fetching a limited number of rows at a time and
therefore not risking exhausting memory.  A disadvantage of the scheme
is that 'aligned' output mode will align each group of rows independently
leading to odd-looking output, but all the other output formats work
reasonably well.  Chris Mair, with some additional hacking by moi.

17 years agoSeparate prepared statement and bind parameters with comma.
Bruce Momjian [Tue, 29 Aug 2006 20:10:42 +0000 (20:10 +0000)]
Separate prepared statement and bind parameters with comma.

Fix printing of NULL bind parameters, use "NULL".

17 years agoAdd autovacuum item:
Bruce Momjian [Tue, 29 Aug 2006 16:22:03 +0000 (16:22 +0000)]
Add autovacuum item:

>  o Turn on by default

17 years agoFix mistyping
Teodor Sigaev [Tue, 29 Aug 2006 15:47:48 +0000 (15:47 +0000)]
Fix mistyping

17 years agoInvent an assign-hook mechanism for psql variables similar to the one
Tom Lane [Tue, 29 Aug 2006 15:19:51 +0000 (15:19 +0000)]
Invent an assign-hook mechanism for psql variables similar to the one
existing for backend GUC variables, and use this to eliminate repeated
fetching/parsing of psql variables in psql's inner loops.  In a trivial
test with lots of 'select 1;' commands, psql's CPU time went down almost
10%, although of course the effect on total elapsed time was much less.
Per discussion about how to ensure the upcoming FETCH_COUNT patch doesn't
cost any performance when not being used.

17 years agoFix BUG #2594: Gin Indexes cause server to crash when it builds on empty table
Teodor Sigaev [Tue, 29 Aug 2006 14:05:44 +0000 (14:05 +0000)]
Fix BUG #2594: Gin Indexes cause server to crash when it builds on empty table

17 years agoAdd description of tsvector type layout
Teodor Sigaev [Tue, 29 Aug 2006 13:57:34 +0000 (13:57 +0000)]
Add description of tsvector type layout

17 years agoRemove pos comparison in silly_cmp_tsvector(): it is not a semantically significant
Teodor Sigaev [Tue, 29 Aug 2006 13:39:20 +0000 (13:39 +0000)]
Remove pos comparison in silly_cmp_tsvector(): it is not a semantically significant

17 years agoFix incorrect length of lexemes in silly_cmp_tsvector()
Teodor Sigaev [Tue, 29 Aug 2006 13:31:54 +0000 (13:31 +0000)]
Fix incorrect length of lexemes in silly_cmp_tsvector()

17 years agoSecond try committing the path changes.
Michael Meskes [Tue, 29 Aug 2006 13:23:27 +0000 (13:23 +0000)]
Second try committing the path changes.

17 years agoRemoved debug option from pg_regress call.
Michael Meskes [Tue, 29 Aug 2006 12:33:45 +0000 (12:33 +0000)]
Removed debug option from pg_regress call.
Changed file paths in expected files too.

17 years agoFixed parser and library to allow empty database names.
Michael Meskes [Tue, 29 Aug 2006 12:24:52 +0000 (12:24 +0000)]
Fixed parser and library to allow empty database names.
Streamlined connection name parsing.
Added Joachim's patch to shorten paths before diffing.

17 years agoRevert change to turn autovacuum on by default.
Peter Eisentraut [Tue, 29 Aug 2006 11:37:47 +0000 (11:37 +0000)]
Revert change to turn autovacuum on by default.

17 years agoOnly call log_after_parse() if necessary.
Bruce Momjian [Tue, 29 Aug 2006 02:32:41 +0000 (02:32 +0000)]
Only call log_after_parse() if necessary.

17 years agoNow bind displays prepare as detail, and execute displays prepare and
Bruce Momjian [Tue, 29 Aug 2006 02:11:30 +0000 (02:11 +0000)]
Now bind displays prepare as detail, and execute displays prepare and
optionally bind.  I re-added the "statement:" label so people will
understand why the line is being printed (it is log_*statement
behavior).

Use single quotes for bind values, instead of double quotes, and double
literal single quotes in bind values (and document that).  I also made
use of the DETAIL line to have much cleaner output.

17 years agoAdd URL:
Bruce Momjian [Mon, 28 Aug 2006 23:22:57 +0000 (23:22 +0000)]
Add URL:

>    http://people.planetpostgresql.org/greg/index.php?/archives/2006/06/10.html
>

17 years agoMove to referential integrity section:
Bruce Momjian [Mon, 28 Aug 2006 23:21:46 +0000 (23:21 +0000)]
Move to referential integrity section:

>  o Allow DEFERRABLE and end-of-statement UNIQUE constraints?
>
>    This would allow UPDATE tab SET col = col + 1 to work if col has
>    a unique index.  Currently, uniqueness checks are done while the
>    command is being executed, rather than at the end of the statement
>    or transaction.
>
<
< * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
<
<   This would allow UPDATE tab SET col = col + 1 to work if col has
<   a unique index.  Currently, uniqueness checks are done while the
<   command is being executed, rather than at the end of the statement
<   or transaction.

17 years agoUpdate:
Bruce Momjian [Mon, 28 Aug 2006 23:20:35 +0000 (23:20 +0000)]
Update:

< * Allow DEFERRABLE UNIQUE constraints?
> * Allow DEFERRABLE and end-of-statement UNIQUE constraints?
>
>   This would allow UPDATE tab SET col = col + 1 to work if col has
>   a unique index.  Currently, uniqueness checks are done while the
>   command is being executed, rather than at the end of the statement
>   or transaction.
>

17 years agoFix pgstat_report_waiting() to not dump core if called before
Tom Lane [Mon, 28 Aug 2006 19:38:09 +0000 (19:38 +0000)]
Fix pgstat_report_waiting() to not dump core if called before
pgstat_bestart() has been called; else any lock-block occurring
during InitPostgres() is disastrous.  I believe this explains
recent wasp regression failure; at least it explains the crash I
got while trying to duplicate the problem.  I also made
pgstat_report_activity() safe against the same scenario, just
in case.  The report_waiting hazard was created by my patch of
19-Aug to include waiting status in pg_stat_activity.

17 years agoPartial fix for ecpg's VPATH problems. It compiles and successfully
Tom Lane [Mon, 28 Aug 2006 16:13:11 +0000 (16:13 +0000)]
Partial fix for ecpg's VPATH problems.  It compiles and successfully
builds all the files needed for its regression tests, but the tests
themselves fail because of diffs in the #line directives output by
ecpg itself.  Not sure what to do about that.

17 years agoTweak trivial_subqueryscan() to consider a SubqueryScan's targetlist
Tom Lane [Mon, 28 Aug 2006 14:32:41 +0000 (14:32 +0000)]
Tweak trivial_subqueryscan() to consider a SubqueryScan's targetlist
trivial if it contains either Vars referencing the corresponding subplan
columns, or Consts equaling the corresponding subplan columns.  This
lets the planner eliminate the SubqueryScan in some cases generated by
generate_setop_tlist().

17 years agoTurn autovacuum on by default. (stats_row_level is also on by default.)
Peter Eisentraut [Mon, 28 Aug 2006 13:37:18 +0000 (13:37 +0000)]
Turn autovacuum on by default.  (stats_row_level is also on by default.)
Threshold and scale factor are cut in half for more aggressive behavior.

17 years agoAdd new return codes SPI_OK_INSERT_RETURNING etc to the SPI API.
Tom Lane [Sun, 27 Aug 2006 23:47:58 +0000 (23:47 +0000)]
Add new return codes SPI_OK_INSERT_RETURNING etc to the SPI API.
Fix all the standard PLs to be able to return tuples from FOO_RETURNING
statements as well as utility statements that return tuples.  Also,
fix oversight that SPI_processed wasn't set for a utility statement
returning tuples.  Per recent discussion.

17 years agoAdd some notes about why it's not a bug that RI_FKey_check calls
Tom Lane [Sun, 27 Aug 2006 21:41:21 +0000 (21:41 +0000)]
Add some notes about why it's not a bug that RI_FKey_check calls
HeapTupleSatisfiesItself without doing LockBuffer first.  This code
is a bit fragile, but AFAICS it's not actually broken.

17 years agoAdd a function GetLockConflicts() to lock.c to report xacts holding
Tom Lane [Sun, 27 Aug 2006 19:14:34 +0000 (19:14 +0000)]
Add a function GetLockConflicts() to lock.c to report xacts holding
locks that would conflict with a specified lock request, without
actually trying to get that lock.  Use this instead of the former ad hoc
method of doing the first wait step in CREATE INDEX CONCURRENTLY.
Fixes problem with undetected deadlock and in many cases will allow the
index creation to proceed sooner than it otherwise could've.  Per
discussion with Greg Stark.

17 years agoMove xact.c's partial support for Lists of TransactionIds into pg_list.h.
Tom Lane [Sun, 27 Aug 2006 19:11:46 +0000 (19:11 +0000)]
Move xact.c's partial support for Lists of TransactionIds into pg_list.h.
Needed because lock.c is now going to use the same type of list.

17 years ago- Enabled single-quoted connection targets.
Michael Meskes [Sun, 27 Aug 2006 16:15:42 +0000 (16:15 +0000)]
- Enabled single-quoted connection targets.
- Fixed a memory leak/segfault in unsuccessful connection.
- Some changes to test files.

17 years agoMake saveHistory work properly on OS X when HISTFILE is set to /dev/null.
Tom Lane [Sun, 27 Aug 2006 15:05:20 +0000 (15:05 +0000)]
Make saveHistory work properly on OS X when HISTFILE is set to /dev/null.
Per discussion with Martin Atukunda.

17 years agoChanged double output.
Michael Meskes [Sat, 26 Aug 2006 11:50:00 +0000 (11:50 +0000)]
Changed double output.

17 years agoAdd blank line.
Bruce Momjian [Fri, 25 Aug 2006 23:45:02 +0000 (23:45 +0000)]
Add blank line.

17 years agoAdd blank line.>
Bruce Momjian [Fri, 25 Aug 2006 23:44:04 +0000 (23:44 +0000)]
Add blank line.>

17 years agoAdd:
Bruce Momjian [Fri, 25 Aug 2006 23:43:46 +0000 (23:43 +0000)]
Add:

<
> * Implement SQL:2003 window functions

17 years agoAdd 'feedback' section heading to xml2.
Bruce Momjian [Fri, 25 Aug 2006 23:43:11 +0000 (23:43 +0000)]
Add 'feedback' section heading to xml2.

17 years agoFix regression tests: after changing comparing function
Teodor Sigaev [Fri, 25 Aug 2006 07:39:08 +0000 (07:39 +0000)]
Fix regression tests: after changing comparing function
order is changed.

17 years agoAdd the ability to create indexes 'concurrently', that is, without
Tom Lane [Fri, 25 Aug 2006 04:06:58 +0000 (04:06 +0000)]
Add the ability to create indexes 'concurrently', that is, without
blocking concurrent writes to the table.  Greg Stark, with a little help
from Tom Lane.

17 years agoFix compare bug for tsvector: problem was in aligment. Per Stefan Kaltenbrunner ...
Teodor Sigaev [Thu, 24 Aug 2006 17:37:34 +0000 (17:37 +0000)]
Fix compare bug for tsvector: problem was in aligment. Per Stefan Kaltenbrunner <stefan@kaltenbrunner.cc> and Phil Frost <indigo@bitglue.com>

17 years agoAdd:
Bruce Momjian [Thu, 24 Aug 2006 16:38:13 +0000 (16:38 +0000)]
Add:

> * Allow inlining of set-returning functions

17 years agoSynced parser
Michael Meskes [Thu, 24 Aug 2006 12:31:33 +0000 (12:31 +0000)]
Synced parser

17 years agoNeeded more stuff from c.h.
Michael Meskes [Thu, 24 Aug 2006 10:48:21 +0000 (10:48 +0000)]
Needed more stuff from c.h.

17 years agoFixed of by one variable size.
Michael Meskes [Thu, 24 Aug 2006 10:35:58 +0000 (10:35 +0000)]
Fixed of by one variable size.

17 years agoNeed more defines in ecpg_config.h, patch sent by Rocco Altier <RoccoA@Routescape...
Michael Meskes [Thu, 24 Aug 2006 09:37:07 +0000 (09:37 +0000)]
Need more defines in ecpg_config.h, patch sent by Rocco Altier <RoccoA@Routescape.com>

17 years agoAdd some debug logging code to AllocateFile's failure path to log the
Tom Lane [Thu, 24 Aug 2006 03:15:43 +0000 (03:15 +0000)]
Add some debug logging code to AllocateFile's failure path to log the
specific Windows error code (GetLastError).  This is a hopefully temporary
hack to try to diagnose rare failures.  Magnus Hagander

17 years agoUpdate XML2 documentation for xpath_table().
Bruce Momjian [Thu, 24 Aug 2006 01:41:06 +0000 (01:41 +0000)]
Update XML2 documentation for xpath_table().

John Gray

17 years agoOptimize the case where a btree indexscan has current and mark positions
Tom Lane [Thu, 24 Aug 2006 01:18:34 +0000 (01:18 +0000)]
Optimize the case where a btree indexscan has current and mark positions
on the same index page; we can avoid data copying as well as buffer refcount
manipulations in this common case.  Makes for a small but noticeable
improvement in mergejoin speed.

Heikki Linnakangas

17 years agoAdd "AS" item:
Bruce Momjian [Wed, 23 Aug 2006 18:32:02 +0000 (18:32 +0000)]
Add "AS" item:

< * All backends running as threads in a single process (not want)
> * All backends running as threads in a single process (not wanted)
< * Optimizer hints (not want)
> * Optimizer hints (not wanted)
>
> * Allow AS in "SELECT col AS label" to be optional (not wanted)
>
>   Because we support postfix operators, it isn't possible to make AS
>   optional and continue to use bison.
>   http://archives.postgresql.org/pgsql-sql/2006-08/msg00164.php

17 years agoSomehow my version compiled but the change didn't make it to CVS.
Michael Meskes [Wed, 23 Aug 2006 13:57:27 +0000 (13:57 +0000)]
Somehow my version compiled but the change didn't make it to CVS.

17 years agoUsed autoconf 2.59 this time.
Michael Meskes [Wed, 23 Aug 2006 12:59:18 +0000 (12:59 +0000)]
Used autoconf 2.59 this time.

17 years agoArgh, forgot the new file yet again.
Michael Meskes [Wed, 23 Aug 2006 12:51:26 +0000 (12:51 +0000)]
Argh, forgot the new file yet again.

17 years agoReplaced double-quote-fix with a hopefully better version.
Michael Meskes [Wed, 23 Aug 2006 12:01:53 +0000 (12:01 +0000)]
Replaced double-quote-fix with a hopefully better version.
Use initializer string length as size for character strings.
Added ecpg_config.h file that is created via configure.

17 years agoUpdate COPY/VIEW item:
Bruce Momjian [Wed, 23 Aug 2006 02:48:12 +0000 (02:48 +0000)]
Update COPY/VIEW item:

<  o Allow COPY to output from SELECT
>  o Allow COPY (SELECT ...) TO 'filename'
<    COPY should also be able to output views.
>    COPY should also be able to output views using COPY (SELECT
>    * FROM view) TO 'filename' internally.

17 years agoDescriptor values were quoted twice.
Michael Meskes [Tue, 22 Aug 2006 12:46:18 +0000 (12:46 +0000)]
Descriptor values were quoted twice.
Fixed some regression test problems.

17 years agoIn new "invalid byte sequence" error hint, call it "error", not
Bruce Momjian [Tue, 22 Aug 2006 12:11:28 +0000 (12:11 +0000)]
In new "invalid byte sequence" error hint, call it "error", not
"failure".

17 years agoAdd info for get_* functions:
Bruce Momjian [Tue, 22 Aug 2006 03:40:39 +0000 (03:40 +0000)]
Add info for get_* functions:

>
>   These would be for application use, not for use by pg_dump.
>

17 years agoAdd hint for "invalid byte sequence for encoding" error message,
Bruce Momjian [Tue, 22 Aug 2006 03:30:20 +0000 (03:30 +0000)]
Add hint for "invalid byte sequence for encoding" error message,
suggesting review of client_encoding.

17 years agoOoops, ldap fix for win32 broke the non-win32 case.
Tom Lane [Tue, 22 Aug 2006 02:23:45 +0000 (02:23 +0000)]
Ooops, ldap fix for win32 broke the non-win32 case.

17 years agoGeneralize width_bucket() documentation because b1/b2 can be
Bruce Momjian [Tue, 22 Aug 2006 00:49:19 +0000 (00:49 +0000)]
Generalize width_bucket() documentation because b1/b2 can be
interchanged.

17 years agoAdd space between Win2000 and SP4.
Bruce Momjian [Tue, 22 Aug 2006 00:45:34 +0000 (00:45 +0000)]
Add space between Win2000 and SP4.

17 years agoFix encrypted-LDAP support so that it doesn't cause the server to fail
Tom Lane [Mon, 21 Aug 2006 19:21:38 +0000 (19:21 +0000)]
Fix encrypted-LDAP support so that it doesn't cause the server to fail
entirely on older Windows platforms without the needed library function.
Magnus Hagander

17 years agoMinor code rearrangement to save a few cycles in RI_FKey_check when
Tom Lane [Mon, 21 Aug 2006 19:15:29 +0000 (19:15 +0000)]
Minor code rearrangement to save a few cycles in RI_FKey_check when
the subject tuple is already deleted: we need not open the pk_rel
until after we check that.

17 years agoRemove obsolete tip about casting bool to int, per Taiki Yamaguchi.
Tom Lane [Mon, 21 Aug 2006 16:23:46 +0000 (16:23 +0000)]
Remove obsolete tip about casting bool to int, per Taiki Yamaguchi.

17 years agoMake the server track an 'XID epoch', that is, maintain higher-order bits
Tom Lane [Mon, 21 Aug 2006 16:16:31 +0000 (16:16 +0000)]
Make the server track an 'XID epoch', that is, maintain higher-order bits
of the transaction ID counter.  Nothing is done with the epoch except to
store it in checkpoint records, but this provides a foundation with which
add-on code can pretend that XIDs never wrap around.  This is a severely
trimmed and rewritten version of the xxid patch submitted by Marko Kreen.
Per discussion, the epoch counter seems the only part of xxid that really
needs to be in the core server.

17 years agoDone:
Bruce Momjian [Mon, 21 Aug 2006 12:50:29 +0000 (12:50 +0000)]
Done:

<  o Add a function to support Parse/DescribeStatement capability
>  o -Add a function to support Parse/DescribeStatement capability

17 years agoRemove items, not needed anymore:
Bruce Momjian [Mon, 21 Aug 2006 12:20:26 +0000 (12:20 +0000)]
Remove items, not needed anymore:

< * %Disallow changing DEFAULT expression of a SERIAL column?
<
<   This should be done only if the existing SERIAL problems cannot be
<   fixed.
<
< * %Disallow ALTER SEQUENCE changes for SERIAL sequences because pg_dump
<   does not dump the changes

17 years agoFixed a few memory leaks. One indeed was part of a loop.
Michael Meskes [Mon, 21 Aug 2006 10:48:21 +0000 (10:48 +0000)]
Fixed a few memory leaks. One indeed was part of a loop.

17 years agoFix all known problems with pg_dump's handling of serial sequences
Tom Lane [Mon, 21 Aug 2006 00:57:26 +0000 (00:57 +0000)]
Fix all known problems with pg_dump's handling of serial sequences
by abandoning the idea that it should say SERIAL in the dump.  Instead,
dump serial sequences and column defaults just like regular ones.
Add a new backend command ALTER SEQUENCE OWNED BY to let pg_dump recreate
the sequence-to-column dependency that was formerly created "behind the
scenes" by SERIAL.  This restores SERIAL to being truly "just a macro"
consisting of component operations that can be stated explicitly in SQL.
Furthermore, the new command allows sequence ownership to be reassigned,
so that old mistakes can be cleaned up.

Also, downgrade the OWNED-BY dependency from INTERNAL to AUTO, since there
is no longer any very compelling argument why the sequence couldn't be
dropped while keeping the column.  (This forces initdb, to be sure the
right kinds of dependencies are in there.)

Along the way, add checks to prevent ALTER OWNER or SET SCHEMA on an
owned sequence; you can now only do this indirectly by changing the
owning table's owner or schema.  This is an oversight in previous
releases, but probably not worth back-patching.

17 years agoFix DROP OWNED BY to correctly consider the implicitly-deleted objects list for
Alvaro Herrera [Sun, 20 Aug 2006 21:56:16 +0000 (21:56 +0000)]
Fix DROP OWNED BY to correctly consider the implicitly-deleted objects list for
each object to be deleted, instead of the previous hack that just skipped
INTERNAL dependencies, which didn't really work.  Per report from Tom Lane.

To do this, introduce a new performMultipleDeletions entry point in
dependency.c to delete multiple objects at once.  The dependency code then has
the responsability of tracking INTERNAL and AUTO dependencies as needed.

Along the way, change ObjectAddresses so that we can allocate an ObjectAddress
list from outside dependency.c and not have to export the internal
representation.

17 years agoFixed a few trivial memory leaks reported by Coverity just to test my setup.
Michael Meskes [Sun, 20 Aug 2006 16:08:09 +0000 (16:08 +0000)]
Fixed a few trivial memory leaks reported by Coverity just to test my setup.

17 years agoAdded another test
Michael Meskes [Sat, 19 Aug 2006 15:10:00 +0000 (15:10 +0000)]
Added another test

17 years agoMore SoC stuff
Michael Meskes [Sat, 19 Aug 2006 13:42:40 +0000 (13:42 +0000)]
More SoC stuff

17 years agoSuppress subquery pullup/pushdown when a subquery contains volatile
Tom Lane [Sat, 19 Aug 2006 02:48:53 +0000 (02:48 +0000)]
Suppress subquery pullup/pushdown when a subquery contains volatile
functions in its targetlist, to avoid introducing multiple evaluations
of volatile functions that textually appear only once.  This is a
slightly tighter version of Jaime Casanova's recent patch.

17 years agoAdd a 'waiting' column to pg_stat_activity to carry the same information
Tom Lane [Sat, 19 Aug 2006 01:36:34 +0000 (01:36 +0000)]
Add a 'waiting' column to pg_stat_activity to carry the same information
that ps_status provides by appending 'waiting' to the PS display.  This
completes the project of making it feasible to turn off process title
updates and instead rely on pg_stat_activity.  Per my suggestion a few
weeks ago.