OSDN Git Service

pg-rex/syncrep.git
18 years agoDefault stats_command_string to 'on', now that its overhead is minimal.
Bruce Momjian [Tue, 27 Jun 2006 19:07:50 +0000 (19:07 +0000)]
Default stats_command_string to 'on', now that its overhead is minimal.

18 years agoPut #ifdef NOT_USED around posix_fadvise call. We may want to resurrect
Tom Lane [Tue, 27 Jun 2006 18:59:17 +0000 (18:59 +0000)]
Put #ifdef NOT_USED around posix_fadvise call.  We may want to resurrect
this someday, but right now it seems that posix_fadvise is immature to
the point of being broken on many platforms ... and we don't have any
benchmark evidence proving it's worth spending time on.

18 years agoRevert patch, doesn't do what it should:
Bruce Momjian [Tue, 27 Jun 2006 18:35:05 +0000 (18:35 +0000)]
Revert patch, doesn't do what it should:

* %Disallow changing default expression of a SERIAL column

Dhanaraj M

18 years agoExtend the MinimalTuple concept to tuplesort.c, thereby reducing the
Tom Lane [Tue, 27 Jun 2006 16:53:02 +0000 (16:53 +0000)]
Extend the MinimalTuple concept to tuplesort.c, thereby reducing the
per-tuple space overhead for sorts in memory.  I chose to replace the
previous patch that tried to write out the bare minimum amount of data
when sorting on disk; instead, just dump the MinimalTuples as-is.  This
wastes 3 to 10 bytes per tuple depending on architecture and null-bitmap
length, but the simplification in the writetup/readtup routines seems
worth it.

18 years agoWill revert in next patch more cleanly.
Bruce Momjian [Tue, 27 Jun 2006 16:42:00 +0000 (16:42 +0000)]
Will revert in next patch more cleanly.

> * -Disallow changing DEFAULT expression of a SERIAL column

18 years agoNot done:
Bruce Momjian [Tue, 27 Jun 2006 16:39:27 +0000 (16:39 +0000)]
Not done:

< * -Disallow changing DEFAULT expression of a SERIAL column
> * %Disallow changing DEFAULT expression of a SERIAL column

18 years agoClamp last_anl_tuples to n_live_tuples, in case we vacuum a table without
Alvaro Herrera [Tue, 27 Jun 2006 03:45:16 +0000 (03:45 +0000)]
Clamp last_anl_tuples to n_live_tuples, in case we vacuum a table without
analyzing, so that future analyze threshold calculations don't get confused.
Also, make sure we correctly track the decrease of live tuples cause by
deletes.

Per report from Dylan Hansen, patches by Tom Lane and me.

18 years agoAdd INCLUDING CONSTRAINTS to CREATE TABLE LIKE.
Bruce Momjian [Tue, 27 Jun 2006 03:43:20 +0000 (03:43 +0000)]
Add INCLUDING CONSTRAINTS to CREATE TABLE LIKE.

Greg Stark

18 years agoDone:
Bruce Momjian [Tue, 27 Jun 2006 03:22:45 +0000 (03:22 +0000)]
Done:

< * %Disallow changing DEFAULT expression of a SERIAL column?
<
<   This should be done only if the existing SERIAL problems cannot be
<   fixed.
<
> * -Disallow changing DEFAULT expression of a SERIAL column

18 years agoDisallow changing/dropping default expression of a SERIAL column
Bruce Momjian [Tue, 27 Jun 2006 03:21:55 +0000 (03:21 +0000)]
Disallow changing/dropping default expression of a SERIAL column

Dhanaraj M

18 years agoFix for recent Win32 pg_dump tar temp file patch.
Bruce Momjian [Tue, 27 Jun 2006 02:56:41 +0000 (02:56 +0000)]
Fix for recent Win32 pg_dump tar temp file patch.

Hiroshi Saito

18 years agoCreate infrastructure for 'MinimalTuple' representation of in-memory
Tom Lane [Tue, 27 Jun 2006 02:51:40 +0000 (02:51 +0000)]
Create infrastructure for 'MinimalTuple' representation of in-memory
tuples with less header overhead than a regular HeapTuple, per my
recent proposal.  Teach TupleTableSlot code how to deal with these.
As proof of concept, change tuplestore.c to store MinimalTuples instead
of HeapTuples.  Future patches will expand the concept to other places
where it is useful.

18 years agoOn Win32, use loop to create pg_dump temporary tar file in the current
Bruce Momjian [Tue, 27 Jun 2006 01:16:58 +0000 (01:16 +0000)]
On Win32, use loop to create pg_dump temporary tar file in the current
directory, not in device root, for permission reasons.

Backpatch to 8.1.X.

18 years agoDone:
Bruce Momjian [Tue, 27 Jun 2006 00:03:44 +0000 (00:03 +0000)]
Done:

>  o -Add PQescapeIdentifier()

18 years agoAdd PQescapeIdentifier() to libpq
Bruce Momjian [Tue, 27 Jun 2006 00:03:42 +0000 (00:03 +0000)]
Add PQescapeIdentifier() to libpq

Christopher Kings-Lynne

18 years agoFix blank line.
Bruce Momjian [Mon, 26 Jun 2006 23:53:14 +0000 (23:53 +0000)]
Fix blank line.

18 years agoAdd comment about the use of pg_config_paths.h.
Bruce Momjian [Mon, 26 Jun 2006 18:40:50 +0000 (18:40 +0000)]
Add comment about the use of pg_config_paths.h.

18 years agoChange the row constructor syntax (ROW(...)) so that list elements foo.*
Tom Lane [Mon, 26 Jun 2006 17:24:41 +0000 (17:24 +0000)]
Change the row constructor syntax (ROW(...)) so that list elements foo.*
will be expanded to a list of their member fields, rather than creating
a nested rowtype field as formerly.  (The old behavior is still available
by omitting '.*'.)  This syntax is not allowed by the SQL spec AFAICS,
so changing its behavior doesn't violate the spec.  The new behavior is
substantially more useful since it allows, for example, triggers to check
for data changes with 'if row(new.*) is distinct from row(old.*)'.  Per
my recent proposal.

18 years agoAdded missing braces to prevent a segfault after usage of an undeclared cursor.
Michael Meskes [Mon, 26 Jun 2006 14:12:02 +0000 (14:12 +0000)]
Added missing braces to prevent a segfault after usage of an undeclared cursor.

18 years agoUse Unix newlines, instead of DOS ones.
Bruce Momjian [Mon, 26 Jun 2006 13:04:05 +0000 (13:04 +0000)]
Use Unix newlines, instead of DOS ones.

18 years agoUse Unix line endings instead of DOS ones, per Magnus.
Bruce Momjian [Mon, 26 Jun 2006 12:59:44 +0000 (12:59 +0000)]
Use Unix line endings instead of DOS ones, per Magnus.

18 years agoUpdate include files to Unix newline format, per report from Magnus.
Bruce Momjian [Mon, 26 Jun 2006 12:52:12 +0000 (12:52 +0000)]
Update include files to Unix newline format, per report from Magnus.

18 years agoUpdate comment description of geo routines and move comment to more
Bruce Momjian [Mon, 26 Jun 2006 12:32:42 +0000 (12:32 +0000)]
Update comment description of geo routines and move comment to more
relevant location.

18 years agoAdded some more coverity report patches send in by Joachim Wieland <joe@mcknight...
Michael Meskes [Mon, 26 Jun 2006 09:20:09 +0000 (09:20 +0000)]
Added some more coverity report patches send in by Joachim Wieland <joe@mcknight.de>.

18 years agoUpdate my info:
Bruce Momjian [Sun, 25 Jun 2006 18:52:03 +0000 (18:52 +0000)]
Update my info:

> * Bruce is Bruce Momjian <bruce@momjian.us> of EnterpriseDB

18 years agoTweak dynahash.c to avoid wasting memory space in non-shared hash tables.
Tom Lane [Sun, 25 Jun 2006 18:29:49 +0000 (18:29 +0000)]
Tweak dynahash.c to avoid wasting memory space in non-shared hash tables.
palloc() will normally round allocation requests up to the next power of 2,
so make dynahash choose allocation sizes that are as close to a power of 2
as possible.
Back-patch to 8.1 --- the problem exists further back, but a much larger
patch would be needed and it doesn't seem worth taking any risks.

18 years agoAdd:
Bruce Momjian [Sun, 25 Jun 2006 16:27:41 +0000 (16:27 +0000)]
Add:

< * Reuse index tuples that point to rows that are not visible to anyone?
> * Reuse index tuples that point to heap tuples that are not visible to
>   anyone?

18 years agoMoved some free() calls that coverity correctly complains about.
Michael Meskes [Sun, 25 Jun 2006 09:38:40 +0000 (09:38 +0000)]
Moved some free() calls that coverity correctly complains about.

18 years agoOur version of getopt_long does not set optarg upon detecting an error, as
Alvaro Herrera [Sun, 25 Jun 2006 04:37:55 +0000 (04:37 +0000)]
Our version of getopt_long does not set optarg upon detecting an error, as
opposed to what other versions apparently do, so it's not safe to print an
error message.  Besides, getopt_long itself already did, so it's redundant
anyway.

18 years agoRemove individual user copyright because the code is contributed to
Bruce Momjian [Sun, 25 Jun 2006 01:45:32 +0000 (01:45 +0000)]
Remove individual user copyright because the code is contributed to
PGDG:

> Yes. In fact the copyright belongs to credativ GmbH the company that
> paid Carsten for his work. As you may or may not know I'm the CEO of
> that company and can assure you that his work was contributed to the
> PostgreSQL project.

18 years agoFix GEVHDRSZ for Win32.
Bruce Momjian [Sun, 25 Jun 2006 01:02:12 +0000 (01:02 +0000)]
Fix GEVHDRSZ for Win32.

Magnus Hagander

18 years agoFix Win32/Cygwin problems:
Bruce Momjian [Sun, 25 Jun 2006 00:18:24 +0000 (00:18 +0000)]
Fix Win32/Cygwin problems:

After updating to the latest cvs, and also building most of the addons
(like PLs), the following patch is neededf for win32 + Visual C++.

* Switch to use the new win32 semaphore code
* Rename win32_open to pgwin32_open. win32_open collides with symbols
defined in Perl. MingW didn't detect ig, MSVC did. And it's a bit too
generic a name to export globally, imho...
* Python defines some partially broken #pragmas in the headers when
doing a debug build. Workaround.

Magnus Hagander

18 years agoUpdate entry:
Bruce Momjian [Sat, 24 Jun 2006 23:47:58 +0000 (23:47 +0000)]
Update entry:

< * Allow heap reuse of UPDATEd rows if old and new versions are on the
<   same heap page?
> * Allow heap reuse of UPDATEd rows if no indexed columns are changed,
>   and old and new versions are on the same heap page?
<   This is possible for same-page updates because a single index row
<   can point to both old and new values.
>   While vacuum handles DELETEs fine, updating of non-indexed columns, like
>   counters, are difficult for VACUUM to handle efficiently.  This method
>   is possible for same-page updates because a single index row can be
>   used to point to both old and new values.

18 years agoAdd UPDATE entry for row reuse.
Bruce Momjian [Sat, 24 Jun 2006 23:45:02 +0000 (23:45 +0000)]
Add UPDATE entry for row reuse.

>
> * Allow heap reuse of UPDATEd rows if old and new versions are on the
>   same heap page?
>
>   This is possible for same-page updates because a single index row
>   can point to both old and new values.
>   http://archives.postgresql.org/pgsql-hackers/2006-06/msg01305.php

18 years agoFix re-inclusion of port header file on Windows, per Magnus.
Tom Lane [Sat, 24 Jun 2006 14:52:48 +0000 (14:52 +0000)]
Fix re-inclusion of port header file on Windows, per Magnus.

18 years agoClarified copyright notice
Michael Meskes [Fri, 23 Jun 2006 14:50:01 +0000 (14:50 +0000)]
Clarified copyright notice

18 years agoRemove dllinit.c; it was only needed for long-obsolete Cygwin versions,
Tom Lane [Thu, 22 Jun 2006 23:50:35 +0000 (23:50 +0000)]
Remove dllinit.c; it was only needed for long-obsolete Cygwin versions,
and there was some question about its copyright status.

18 years agopg_stop_backup was calling XLogArchiveNotify() twice for the newly created
Tom Lane [Thu, 22 Jun 2006 20:42:57 +0000 (20:42 +0000)]
pg_stop_backup was calling XLogArchiveNotify() twice for the newly created
backup history file.  Bug introduced by the 8.1 change to make pg_stop_backup
delete older history files.  Per report from Masao Fujii.

18 years agoStandalone backends need pgstat_bestart() too, per Qingqing Zhou.
Tom Lane [Thu, 22 Jun 2006 14:36:12 +0000 (14:36 +0000)]
Standalone backends need pgstat_bestart() too, per Qingqing Zhou.

18 years agoMove setup_cancel_handler() up near start of psql main(), where the
Tom Lane [Wed, 21 Jun 2006 19:40:31 +0000 (19:40 +0000)]
Move setup_cancel_handler() up near start of psql main(), where the
setup_win32_locks() call formerly was, to ensure that cancelConnLock is
valid when it needs to be.  Per Yoshiyuki Asaba.

18 years agoRemove ancient kluge that kept nodeAgg.c from crashing on UPDATEs involving
Tom Lane [Wed, 21 Jun 2006 18:39:42 +0000 (18:39 +0000)]
Remove ancient kluge that kept nodeAgg.c from crashing on UPDATEs involving
aggregates.  We just disallowed that, and AFAICS there should be no other
cases where direct (non-aggregated) references to input columns are allowed
in a query with aggregation and no GROUP BY.

18 years agoDisallow aggregate functions in UPDATE commands (unless within a sub-SELECT).
Tom Lane [Wed, 21 Jun 2006 18:30:11 +0000 (18:30 +0000)]
Disallow aggregate functions in UPDATE commands (unless within a sub-SELECT).
This is disallowed by the SQL spec because it doesn't have any very sensible
interpretation.  Historically Postgres has allowed it but behaved strangely.
As of PG 8.1 a server crash is possible if the MIN/MAX index optimization gets
applied; rather than try to "fix" that, it seems best to just enforce the
spec restriction.  Per report from Josh Drake and Alvaro Herrera.

18 years agoEliminate a gratuitously different wording of the 'cannot use aggregate function...
Tom Lane [Wed, 21 Jun 2006 18:09:53 +0000 (18:09 +0000)]
Eliminate a gratuitously different wording of the 'cannot use aggregate function in check constraint' error message.

18 years ago- During dblink_open, if transaction state was IDLE, force cursor count to
Joe Conway [Wed, 21 Jun 2006 16:43:11 +0000 (16:43 +0000)]
- During dblink_open, if transaction state was IDLE, force cursor count to
  initially be 0. This is needed as a previous ABORT might have wiped out
  an automatically opened transaction without maintaining the cursor count.
- Fix regression test expected file for the correct ERROR message, which
  we now get given the above bug fix.

18 years agoClean up psql variable code a little: eliminate unnecessary tests in
Tom Lane [Wed, 21 Jun 2006 16:05:11 +0000 (16:05 +0000)]
Clean up psql variable code a little: eliminate unnecessary tests in
GetVariable() and be consistent about treatment of the list header.
Motivated by noticing strspn() taking an unreasonable percentage of
runtime --- the call removed from GetVariable() was the only one that
could be in a high-usage path ...

18 years agoAdded some more coverity report patches send in by Martijn van Oosterhout <kleptog...
Michael Meskes [Wed, 21 Jun 2006 11:38:07 +0000 (11:38 +0000)]
Added some more coverity report patches send in by Martijn van Oosterhout <kleptog@svana.org>.

18 years agoFixed small typo in changelog
Michael Meskes [Wed, 21 Jun 2006 10:34:48 +0000 (10:34 +0000)]
Fixed small typo in changelog

18 years agoAdded fixed from the coverity report send in by Joachim Wieland <joe@mcknight.de>
Michael Meskes [Wed, 21 Jun 2006 10:24:41 +0000 (10:24 +0000)]
Added fixed from the coverity report send in by Joachim Wieland <joe@mcknight.de>
Added missing error handling in a few functions in ecpglib

18 years agoRemove redundant gettimeofday() calls to the extent practical without
Tom Lane [Tue, 20 Jun 2006 22:52:00 +0000 (22:52 +0000)]
Remove redundant gettimeofday() calls to the extent practical without
changing semantics too much.  statement_timestamp is now set immediately
upon receipt of a client command message, and the various places that used
to do their own gettimeofday() calls to mark command startup are referenced
to that instead.  I have also made stats_command_string use that same
value for pg_stat_activity.query_start for both the command itself and
its eventual replacement by <IDLE> or <idle in transaction>.  There was
some debate about that, but no argument that seemed convincing enough to
justify an extra gettimeofday() call.

18 years agoSplit definitions for md5.c out of crypt.h and into their own header
Tom Lane [Tue, 20 Jun 2006 19:56:52 +0000 (19:56 +0000)]
Split definitions for md5.c out of crypt.h and into their own header
libpq/md5.h, so that there's a clear separation between backend-only
definitions and shared frontend/backend definitions.  (Turns out this
is reversing a bad decision from some years ago...)  Fix up references
to crypt.h as needed.  I looked into moving the code into src/port, but
the headers in src/include/libpq are sufficiently intertwined that it
seems more work than it's worth to do that.

18 years agoSome editorial work on the documentation of the current-date/time
Tom Lane [Mon, 19 Jun 2006 16:13:01 +0000 (16:13 +0000)]
Some editorial work on the documentation of the current-date/time
functions.

18 years agoDo not use already free'ed errmsg, bug found by Joachim Wieland
Michael Meskes [Mon, 19 Jun 2006 09:19:49 +0000 (09:19 +0000)]
Do not use already free'ed errmsg, bug found by Joachim Wieland
<joachim.wieland@credativ.de>

18 years agoTake the statistics collector out of the loop for monitoring backends'
Tom Lane [Mon, 19 Jun 2006 01:51:22 +0000 (01:51 +0000)]
Take the statistics collector out of the loop for monitoring backends'
current commands; instead, store current-status information in shared
memory.  This substantially reduces the overhead of stats_command_string
and also ensures that pg_stat_activity is fully up to date at all times.
Per my recent proposal.

18 years agoUpdate my email address.
Bruce Momjian [Sun, 18 Jun 2006 19:34:00 +0000 (19:34 +0000)]
Update my email address.

18 years agoDon't try to call posix_fadvise() unless <fcntl.h> supplies a declaration
Tom Lane [Sun, 18 Jun 2006 18:30:21 +0000 (18:30 +0000)]
Don't try to call posix_fadvise() unless <fcntl.h> supplies a declaration
for it.  Hopefully will fix core dump evidenced by some buildfarm members
since fadvise patch went in.  The actual definition of the function is not
ABI-compatible with compiler's default assumption in the absence of any
declaration, so it's clearly unsafe to try to call it without seeing a
declaration.

18 years agoFix saveHistory() to not emit bogus complaint during psql exit when using
Tom Lane [Sun, 18 Jun 2006 17:30:48 +0000 (17:30 +0000)]
Fix saveHistory() to not emit bogus complaint during psql exit when using
Darwin's libedit; per my proposal last August.  Also, suppress cast-away-
const warnings.

18 years agoIncrease timeout in statement_timeout test from 1 second to 2 seconds.
Tom Lane [Sun, 18 Jun 2006 16:21:23 +0000 (16:21 +0000)]
Increase timeout in statement_timeout test from 1 second to 2 seconds.
We have once or twice seen failures suggesting that control didn't get
to the exception block before the timeout elapsed, which is unlikely
but not impossible in a parallel regression test (with a dozen other
backends competing for cycles).  This change doesn't completely prevent
the problem of course, but it should reduce the probability enough that
we don't see it anymore.  Per buildfarm results.

18 years agoMerge postmaster and postgres command into just postgres. postmaster
Peter Eisentraut [Sun, 18 Jun 2006 15:38:37 +0000 (15:38 +0000)]
Merge postmaster and postgres command into just postgres.  postmaster
symlink is kept for now for compatibility.  To call single-user mode, use
postgres --single.

18 years agoFix a couple of obvious problems in DROP IF EXISTS patch.
Tom Lane [Fri, 16 Jun 2006 23:50:48 +0000 (23:50 +0000)]
Fix a couple of obvious problems in DROP IF EXISTS patch.

18 years agoCode review for SELECT INTO STRICT patch: use saner choices of error
Tom Lane [Fri, 16 Jun 2006 23:29:27 +0000 (23:29 +0000)]
Code review for SELECT INTO STRICT patch: use saner choices of error
SQLSTATEs, fix some documentation problems.

18 years agoClean up after someone's curious idea that it'd be good to strip
Tom Lane [Fri, 16 Jun 2006 22:41:45 +0000 (22:41 +0000)]
Clean up after someone's curious idea that it'd be good to strip
leading zeroes from the SQLSTATE codes.  They're strings, people,
not numbers.

18 years agodocs for DROP ... IF EXISTS for the following cases:
Andrew Dunstan [Fri, 16 Jun 2006 22:27:55 +0000 (22:27 +0000)]
docs for DROP ... IF EXISTS for the following cases:
  language, tablespace, trigger, rule, opclass, function, aggregate. operator, and cast.

18 years agoAdd:
Bruce Momjian [Fri, 16 Jun 2006 22:08:46 +0000 (22:08 +0000)]
Add:

>  o Allow PL/python to composite types and result sets
>    once buggy assert-enabled versions of python can be detected
>
>    http://archives.postgresql.org/pgsql-patches/2006-04/msg00087.php

18 years agoAdd:
Bruce Momjian [Fri, 16 Jun 2006 22:05:01 +0000 (22:05 +0000)]
Add:

>
> * Fix CREATE CAST on DOMAINs
>
>   http://archives.postgresql.org/pgsql-hackers/2006-05/msg00072.php

18 years agoDocument issues with non-default tablespaces and pg_dumpall restores.
Bruce Momjian [Fri, 16 Jun 2006 22:01:17 +0000 (22:01 +0000)]
Document issues with non-default tablespaces and pg_dumpall restores.

Backpatch documentation addition to 8.1.X.

18 years agoAdd URL for ISO dates:
Bruce Momjian [Fri, 16 Jun 2006 21:41:24 +0000 (21:41 +0000)]
Add URL for ISO dates:

>    http://archives.postgresql.org/pgsql-bugs/2006-04/msg00248.php

18 years agoDROP ... IF EXISTS for the following cases:
Andrew Dunstan [Fri, 16 Jun 2006 20:23:45 +0000 (20:23 +0000)]
DROP ... IF EXISTS for the following cases:
  language, tablespace, trigger, rule, opclass, function, aggregate. operator, and cast.

18 years agoAdd URL for currency type:
Bruce Momjian [Fri, 16 Jun 2006 19:49:50 +0000 (19:49 +0000)]
Add URL for currency type:

>   http://archives.postgresql.org/pgsql-general/2005-08/msg01432.php

18 years agoAdd URL for updatable views:
Bruce Momjian [Fri, 16 Jun 2006 18:50:24 +0000 (18:50 +0000)]
Add URL for updatable views:

>   http://archives.postgresql.org/pgsql-hackers/2006-03/msg00586.php

18 years agoAdd:
Bruce Momjian [Fri, 16 Jun 2006 18:48:42 +0000 (18:48 +0000)]
Add:

> * Consider shrinking expired tuples to just their headers
>   http://archives.postgresql.org/pgsql-patches/2006-03/msg00142.php

18 years agoFix problems with cached tuple descriptors disappearing while still in use
Tom Lane [Fri, 16 Jun 2006 18:42:24 +0000 (18:42 +0000)]
Fix problems with cached tuple descriptors disappearing while still in use
by creating a reference-count mechanism, similar to what we did a long time
ago for catcache entries.  The back branches have an ugly solution involving
lots of extra copies, but this way is more efficient.  Reference counting is
only applied to tupdescs that are actually in caches --- there seems no need
to use it for tupdescs that are generated in the executor, since they'll go
away during plan shutdown by virtue of being in the per-query memory context.
Neil Conway and Tom Lane

18 years agoAdd URL to:
Bruce Momjian [Fri, 16 Jun 2006 18:38:39 +0000 (18:38 +0000)]
Add URL to:

* Auto-fill the free space map by scanning the buffer cache or by
  checking pages written by the background writer

18 years agoAdd URL for freespace map filling.
Bruce Momjian [Fri, 16 Jun 2006 18:37:31 +0000 (18:37 +0000)]
Add URL for freespace map filling.

>
>   http://archives.postgresql.org/pgsql-hackers/2006-02/msg01125.php
>

18 years ago>
Bruce Momjian [Fri, 16 Jun 2006 18:32:07 +0000 (18:32 +0000)]
>
> * Add support for public SYNONYMS
>
>   http://archives.postgresql.org/pgsql-hackers/2006-03/msg00519.php

18 years ago> o Add PL/Python tracebacks
Bruce Momjian [Fri, 16 Jun 2006 18:11:03 +0000 (18:11 +0000)]
>               o Add PL/Python tracebacks
>
>                http://archives.postgresql.org/pgsql-patches/2006-02/msg00288.php

18 years agoAdd:
Bruce Momjian [Fri, 16 Jun 2006 17:44:43 +0000 (17:44 +0000)]
Add:

>  o Fix problems with RETURN NEXT on tables with
>    dropped/added columns after function creation
>
>    http://archives.postgresql.org/pgsql-patches/2006-02/msg00165.php

18 years agoAdd URL for:
Bruce Momjian [Fri, 16 Jun 2006 17:25:38 +0000 (17:25 +0000)]
Add URL for:

> * Allow protocol-level BIND parameter values to be logged
>
>   http://archives.postgresql.org/pgsql-hackers/2006-02/msg00165.php

18 years agoAdd to locale TODO.detail.
Bruce Momjian [Fri, 16 Jun 2006 17:14:32 +0000 (17:14 +0000)]
Add to locale TODO.detail.

18 years agoAdd URL for ISO date/time syntax:
Bruce Momjian [Fri, 16 Jun 2006 16:59:30 +0000 (16:59 +0000)]
Add URL for ISO date/time syntax:
>
>    http://archives.postgresql.org/pgsql-hackers/2006-01/msg00250.php
>

18 years agoAdd LDAP documentation missed in code patch.
Bruce Momjian [Fri, 16 Jun 2006 15:16:16 +0000 (15:16 +0000)]
Add LDAP documentation missed in code patch.

Magnus Hagander

18 years agoUpdate preload_library magic block wording.
Bruce Momjian [Fri, 16 Jun 2006 12:47:49 +0000 (12:47 +0000)]
Update preload_library magic block wording.

18 years agoTest for POSIX_FADV_DONTNEED to use posix_fadvise().
Bruce Momjian [Fri, 16 Jun 2006 04:11:48 +0000 (04:11 +0000)]
Test for POSIX_FADV_DONTNEED to use posix_fadvise().

18 years agoFix:
Bruce Momjian [Fri, 16 Jun 2006 04:04:52 +0000 (04:04 +0000)]
Fix:

<    get_employee_salary(emp_id AS 12345, tax_year AS 2001)
>    get_employee_salary(12345 AS emp_id, 2001 AS tax_year)

18 years agoUpdate magic block wording for preloaded libraries.
Bruce Momjian [Fri, 16 Jun 2006 04:02:35 +0000 (04:02 +0000)]
Update magic block wording for preloaded libraries.

18 years agoUse posix_fadvise() to avoid kernel caching of WAL contents on WAL file
Bruce Momjian [Thu, 15 Jun 2006 19:15:00 +0000 (19:15 +0000)]
Use posix_fadvise() to avoid kernel caching of WAL contents on WAL file
close.

ITAGAKI Takahiro

18 years agoAdd STRICT to PL/pgSQL SELECT INTO, so exceptions are thrown if more or
Bruce Momjian [Thu, 15 Jun 2006 18:02:22 +0000 (18:02 +0000)]
Add STRICT to PL/pgSQL SELECT INTO, so exceptions are thrown if more or
less than one row is returned by the SELECT, for Oracle PL/SQL
compatibility.

Improve SELECT INTO documentation.

Matt Miller

18 years agoFix to_ascii() markup error.
Bruce Momjian [Thu, 15 Jun 2006 17:52:48 +0000 (17:52 +0000)]
Fix to_ascii() markup error.

18 years agoAdd mention that preload_libraries check the magic block, so
Bruce Momjian [Thu, 15 Jun 2006 17:49:09 +0000 (17:49 +0000)]
Add mention that preload_libraries check the magic block, so
non-PostgreSQL libraries cannot be loaded using this capability.

18 years agoUpdate syntax suggestion:
Bruce Momjian [Thu, 15 Jun 2006 17:42:02 +0000 (17:42 +0000)]
Update syntax suggestion:

<    get_employee_salary(emp_id => 12345, tax_year => 2001)
>    get_employee_salary(emp_id AS 12345, tax_year AS 2001)

18 years agoMove list of supported to_ascii() encodings from footnote to main
Bruce Momjian [Thu, 15 Jun 2006 17:38:26 +0000 (17:38 +0000)]
Move list of supported to_ascii() encodings from footnote to main
description.

Nis Jorgensen

18 years agoAdd:
Bruce Momjian [Thu, 15 Jun 2006 17:17:41 +0000 (17:17 +0000)]
Add:

>
>  o Fix memory leak from exceptions
>
>    http://archives.postgresql.org/pgsql-performance/2006-06/msg00305.php

18 years agoClarify dynamic pl/pgsql item and add URLs. Restructure server-side
Bruce Momjian [Thu, 15 Jun 2006 16:54:13 +0000 (16:54 +0000)]
Clarify dynamic pl/pgsql item and add URLs.  Restructure server-side
section into PL/pgSQL and non-PL/pgSQL sections:

<  o Fix PL/pgSQL RENAME to work on variables other than OLD/NEW
<  o Allow function parameters to be passed by name,
<    get_employee_salary(emp_id => 12345, tax_year => 2001)
<  o Add Oracle-style packages
<  o Add table function support to pltcl, plpython
<  o Add capability to create and call PROCEDURES
<  o Allow PL/pgSQL to handle %TYPE arrays, e.g. tab.col%TYPE[]
<  o Allow function argument names to be statements from PL/PgSQL
<  o Add MOVE to PL/pgSQL
<  o Add support for polymorphic arguments and return types to
<    languages other than PL/PgSQL
<  o Add support for OUT and INOUT parameters to languages other
<    than PL/PgSQL
<  o Add single-step debugging of PL/PgSQL functions
<  o Allow PL/PgSQL to support WITH HOLD cursors
<  o Allow PL/PgSQL RETURN to return row or record functions
<
<    http://archives.postgresql.org/pgsql-patches/2005-11/msg00045.php
>  o PL/pgSQL
>  o Fix RENAME to work on variables other than OLD/NEW
>  o Allow function parameters to be passed by name,
>    get_employee_salary(emp_id => 12345, tax_year => 2001)
>  o Add Oracle-style packages
>  o Allow handling of %TYPE arrays, e.g. tab.col%TYPE[]
>  o Allow listing of record column names, and access to
>    record columns via variables, e.g. columns := r.(*),
>    tval2 := r.(colname)
>
>    http://archives.postgresql.org/pgsql-patches/2005-07/msg00458.php
>    http://archives.postgresql.org/pgsql-patches/2006-05/msg00302.php
>    http://archives.postgresql.org/pgsql-patches/2006-06/msg00031.php
>
>  o Add MOVE
>  o Add single-step debugging of functions
>  o Add support for WITH HOLD cursors
>  o Allow PL/RETURN to return row or record functions
>
>    http://archives.postgresql.org/pgsql-patches/2005-11/msg00045.php
>
>
>  o Other
>  o Add table function support to pltcl, plpython
>  o Add support for polymorphic arguments and return types to
>    languages other than PL/PgSQL
>  o Add capability to create and call PROCEDURES
>  o Add support for OUT and INOUT parameters to languages other
>    than PL/PgSQL

18 years agoAdd:
Bruce Momjian [Thu, 15 Jun 2006 16:02:32 +0000 (16:02 +0000)]
Add:

>  o Allow PL/PgSQL RETURN to return row or record functions
>
>    http://archives.postgresql.org/pgsql-patches/2005-11/msg00045.php

18 years agoDone, removed:
Bruce Momjian [Thu, 15 Jun 2006 02:23:13 +0000 (02:23 +0000)]
Done, removed:

<
< * Consider padding the spinlock structure to have spinlock registers
<   on different CPU cache line
<
<   http://archives.postgresql.org/pgsql-hackers/2005-10/msg00498.php

18 years agoRemove the limit on the number of entries allowed in catcaches, and
Tom Lane [Thu, 15 Jun 2006 02:08:09 +0000 (02:08 +0000)]
Remove the limit on the number of entries allowed in catcaches, and
remove the infrastructure needed to enforce the limit, ie, the global
LRU list of cache entries.  On small-to-middling databases this wins
because maintaining the LRU list is a waste of time.  On large databases
this wins because it's better to keep more cache entries (we assume
such users can afford to use some more per-backend memory than was
contemplated in the Berkeley-era catcache design).  This provides a
noticeable improvement in the speed of psql \d on a 10000-table
database, though it doesn't make it instantaneous.

While at it, use per-catcache settings for the number of hash buckets
per catcache, rather than the former one-size-fits-all value.  It's a
bit silly to be using the same number of hash buckets for, eg, pg_am
and pg_attribute.  The specific values I used might need some tuning,
but they seem to be in the right ballpark based on CATCACHE_STATS
results from the standard regression tests.

18 years agoAdd to locale TODO.detail.
Bruce Momjian [Wed, 14 Jun 2006 23:28:22 +0000 (23:28 +0000)]
Add to locale TODO.detail.

18 years agoRemove, done:
Bruce Momjian [Wed, 14 Jun 2006 23:19:04 +0000 (23:19 +0000)]
Remove, done:

<
< * Add support for WIN1253 (Greek) encoding
<
<   http://archives.postgresql.org/pgsql-bugs/2005-12/msg00000.php

18 years agoAdd:
Bruce Momjian [Wed, 14 Jun 2006 23:07:55 +0000 (23:07 +0000)]
Add:

>
> * Add support for WIN1253 (Greek) encoding
>
>   http://archives.postgresql.org/pgsql-bugs/2005-12/msg00000.php

18 years agoAdd URL to ILIKE encoding fix:
Bruce Momjian [Wed, 14 Jun 2006 23:03:12 +0000 (23:03 +0000)]
Add URL to ILIKE encoding fix:

>   http://archives.postgresql.org/pgsql-patches/2005-11/msg00173.php

18 years agoAdd:
Bruce Momjian [Wed, 14 Jun 2006 22:31:53 +0000 (22:31 +0000)]
Add:

>
> * Consider increasing internal areas when shared buffers is increased
>
>   http://archives.postgresql.org/pgsql-hackers/2005-10/msg01419.php