]> granicus.if.org Git - postgresql/log
postgresql
19 years agoTest for POSIX.1c 2-arg sigwait() and fail on single-arg version.
Bruce Momjian [Sun, 19 Dec 2004 04:06:23 +0000 (04:06 +0000)]
Test for POSIX.1c 2-arg sigwait() and fail on single-arg version.

19 years agoAdd Win32 version stamps that increment each day for proper SYSTEM32
Bruce Momjian [Sun, 19 Dec 2004 02:16:31 +0000 (02:16 +0000)]
Add Win32 version stamps that increment each day for proper SYSTEM32
DLL pginstaller installs.

19 years agoAllow NetBSD, m64k to compile the ASM spinlock code.
Bruce Momjian [Sat, 18 Dec 2004 22:12:52 +0000 (22:12 +0000)]
Allow NetBSD, m64k to compile the ASM spinlock code.

R?mi Zara

19 years agoAdd note warning that OS X updates usually overwrite /etc/rc.
Tom Lane [Sat, 18 Dec 2004 18:36:33 +0000 (18:36 +0000)]
Add note warning that OS X updates usually overwrite /etc/rc.

19 years agoChanges the doc's to reflect what the user will find as default.
Bruce Momjian [Sat, 18 Dec 2004 04:21:43 +0000 (04:21 +0000)]
Changes the doc's to reflect what the user will find as default.

Robert Treat

19 years agoMany thanks to Joachim Wieland for a big list of corrections / updates.
Bruce Momjian [Sat, 18 Dec 2004 04:18:21 +0000 (04:18 +0000)]
Many thanks to Joachim Wieland for a big list of corrections / updates.

Ian Barwick

19 years agoUpdate from Ian Barwick
Bruce Momjian [Sat, 18 Dec 2004 04:13:20 +0000 (04:13 +0000)]
Update from Ian Barwick

19 years agoMake array_cat more paranoid about checking datatypes in empty arrays.
Tom Lane [Fri, 17 Dec 2004 20:59:58 +0000 (20:59 +0000)]
Make array_cat more paranoid about checking datatypes in empty arrays.

19 years agoarray_map failed to insert correct result type in an empty array.
Tom Lane [Fri, 17 Dec 2004 20:58:26 +0000 (20:58 +0000)]
array_map failed to insert correct result type in an empty array.
Per example from Florian Pflug.

19 years agoMinor copy-editing in tutorial.
Tom Lane [Fri, 17 Dec 2004 04:50:32 +0000 (04:50 +0000)]
Minor copy-editing in tutorial.

19 years agoIn a PGXS build, expect to find the postgres executable already installed,
Tom Lane [Fri, 17 Dec 2004 03:52:49 +0000 (03:52 +0000)]
In a PGXS build, expect to find the postgres executable already installed,
rather than in $(top_builddir)/src/backend/postgres.  Sean Chittenden

19 years agoAllow 'ELSEIF' as an alternative to 'ELSIF' in PL/PgSQL. There have been
Neil Conway [Fri, 17 Dec 2004 03:51:36 +0000 (03:51 +0000)]
Allow 'ELSEIF' as an alternative to 'ELSIF' in PL/PgSQL. There have been
several reports of users being confused when they attempt to use ELSEIF
and run into trouble due to PL/PgSQL's lax parser. The parser will be
improved for 8.1, but we can fix most of the problem by allowing ELSEIF
for now.

19 years agoHook up the plpython result-object nrows and status methods correctly.
Tom Lane [Fri, 17 Dec 2004 02:14:48 +0000 (02:14 +0000)]
Hook up the plpython result-object nrows and status methods correctly.
Adjust documentation to match current reality.

19 years agoFix is-it-time-for-a-checkpoint logic so that checkpoint_segments can
Tom Lane [Fri, 17 Dec 2004 00:10:36 +0000 (00:10 +0000)]
Fix is-it-time-for-a-checkpoint logic so that checkpoint_segments can
usefully be larger than 255.  Per gripe from Simon Riggs.

19 years agoAdd Win32:
Bruce Momjian [Fri, 17 Dec 2004 00:04:30 +0000 (00:04 +0000)]
Add Win32:

>         o Disallow encodings like UTF8 which which PostgreSQL supports
>           but the operating system does not (already disallowed by
>    pginstaller)

19 years agoUpdate release changes list to mention pg_config.h.win32.
Bruce Momjian [Thu, 16 Dec 2004 22:11:06 +0000 (22:11 +0000)]
Update release changes list to mention pg_config.h.win32.

19 years agoUpdate version for Win32-client include file.
Bruce Momjian [Thu, 16 Dec 2004 22:05:07 +0000 (22:05 +0000)]
Update version for Win32-client include file.

19 years agoCause configure --with-tcl to check for presence of <tcl.h>, as per
Tom Lane [Thu, 16 Dec 2004 20:41:01 +0000 (20:41 +0000)]
Cause configure --with-tcl to check for presence of <tcl.h>, as per
gripe from John Gray.  Also fix thinko in pltcl Makefile: if a special
Tcl include directory is specified, that ought to be searched first.

19 years agoAllow --enable-thread-safety with --disable-shared libraries by passing
Bruce Momjian [Thu, 16 Dec 2004 18:13:07 +0000 (18:13 +0000)]
Allow --enable-thread-safety with --disable-shared libraries by passing
PTHREAD_LIBS to all libpq-usage builds.

19 years agoAllow AIX to use --enable-thread-safety by passing PTHREAD_LIBS to
Bruce Momjian [Thu, 16 Dec 2004 17:48:29 +0000 (17:48 +0000)]
Allow AIX to use --enable-thread-safety by passing PTHREAD_LIBS to
binary compiles, and adjust configure tests for AIX.

19 years agoTranslation updates
Peter Eisentraut [Thu, 16 Dec 2004 11:31:55 +0000 (11:31 +0000)]
Translation updates

19 years agoAdd port report
Peter Eisentraut [Thu, 16 Dec 2004 11:23:16 +0000 (11:23 +0000)]
Add port report

19 years agoAdd French IRC channel.
Bruce Momjian [Thu, 16 Dec 2004 01:08:23 +0000 (01:08 +0000)]
Add French IRC channel.

19 years agoAdd Spanish IRC channel to FAQ.
Bruce Momjian [Thu, 16 Dec 2004 00:01:33 +0000 (00:01 +0000)]
Add Spanish IRC channel to FAQ.

19 years agoFix another place broken by new List implementation :-(. Per example
Tom Lane [Wed, 15 Dec 2004 21:13:34 +0000 (21:13 +0000)]
Fix another place broken by new List implementation :-(.  Per example
from goranpop@nspoint.net.  I think this escaped notice because in
simple cases the list is NIL on entry.

19 years agoDisallow SETOF in the input of parseTypeString(). Formerly it was
Tom Lane [Wed, 15 Dec 2004 20:15:17 +0000 (20:15 +0000)]
Disallow SETOF in the input of parseTypeString().  Formerly it was
silently ignored, allowing one to write bizarre things like
DECLARE x setof int;
in plpgsql.  This has misled at least one novice into thinking that
plpgsql variables could be sets ...

19 years agoCalculation of keys_are_unique flag was wrong for cases involving
Tom Lane [Wed, 15 Dec 2004 19:16:39 +0000 (19:16 +0000)]
Calculation of keys_are_unique flag was wrong for cases involving
redundant cross-datatype comparisons.  Per example from Merlin Moncure.

19 years agoNew translations
Peter Eisentraut [Wed, 15 Dec 2004 17:50:20 +0000 (17:50 +0000)]
New translations

19 years agoTranslation updates
Peter Eisentraut [Wed, 15 Dec 2004 17:39:54 +0000 (17:39 +0000)]
Translation updates

19 years agoTranslation update
Peter Eisentraut [Wed, 15 Dec 2004 17:14:11 +0000 (17:14 +0000)]
Translation update

19 years agoUpdate IRC mention.
Bruce Momjian [Wed, 15 Dec 2004 03:11:26 +0000 (03:11 +0000)]
Update IRC mention.

19 years agoIn 'make installcheck' for contrib/, insert 'sleep 1' between subdirectory
Tom Lane [Tue, 14 Dec 2004 22:50:23 +0000 (22:50 +0000)]
In 'make installcheck' for contrib/, insert 'sleep 1' between subdirectory
checks, to ensure the previous test backend has time to quit out of the
regression database.  Also, allow all the checks to be run even if one
of them fails.  Per suggestions from Andrew Dunstan to improve the
usefulness of buildfarm testing.

19 years agoCope with circularities involving a view's ON SELECT rule. I originally
Tom Lane [Tue, 14 Dec 2004 22:16:32 +0000 (22:16 +0000)]
Cope with circularities involving a view's ON SELECT rule.  I originally
thought there couldn't be any, but the folly of this was exposed by an
example from andrew@supernews.com 5-Dec-2004.  The patch applies the
identical logic already used for table constraints and defaults to ON
SELECT rules, so I have reasonable confidence in it even though it might
look like complicated logic.

19 years agoFix pg_dump to ensure that a comment on a table CHECK constraint cannot
Tom Lane [Tue, 14 Dec 2004 21:35:20 +0000 (21:35 +0000)]
Fix pg_dump to ensure that a comment on a table CHECK constraint cannot
be emitted too soon.  The previous code got this right in the case where
the CHECK was emitted as a separate ALTER TABLE command, but not in the
case where the CHECK is emitted right in CREATE TABLE.  Per report from
Slawomir Sudnik.

Note: this code is pretty ugly; it'd perhaps be better to treat comments
as independently sortable dump objects.  That'd be much too invasive a
change for RC time though.

19 years agoUpdate aix cc_r wording.
Bruce Momjian [Tue, 14 Dec 2004 14:53:53 +0000 (14:53 +0000)]
Update aix cc_r wording.

19 years agoMention aix cc_r is not supported, and why
Bruce Momjian [Tue, 14 Dec 2004 12:58:29 +0000 (12:58 +0000)]
Mention aix cc_r is not supported, and why

19 years agoFix typo
Peter Eisentraut [Tue, 14 Dec 2004 11:11:20 +0000 (11:11 +0000)]
Fix typo

19 years agoAdd port reports
Peter Eisentraut [Tue, 14 Dec 2004 10:24:27 +0000 (10:24 +0000)]
Add port reports

19 years agoPrevent pg_resetxlog from being run as root. If this is allowed, some
Neil Conway [Tue, 14 Dec 2004 01:59:41 +0000 (01:59 +0000)]
Prevent pg_resetxlog from being run as root. If this is allowed, some
root-owned files will be written to the data directory, leaving it in an
unusable state.

19 years agoAdd more port reports.
Peter Eisentraut [Tue, 14 Dec 2004 00:36:12 +0000 (00:36 +0000)]
Add more port reports.

19 years agoNew translation
Peter Eisentraut [Mon, 13 Dec 2004 22:54:20 +0000 (22:54 +0000)]
New translation

19 years agoTranslation updates
Peter Eisentraut [Mon, 13 Dec 2004 21:49:10 +0000 (21:49 +0000)]
Translation updates

19 years agoRemove:
Bruce Momjian [Mon, 13 Dec 2004 18:40:30 +0000 (18:40 +0000)]
Remove:

<  o Allow the shared memory address to be configured via GUC

19 years agoSpell checker run
Peter Eisentraut [Mon, 13 Dec 2004 18:05:10 +0000 (18:05 +0000)]
Spell checker run

19 years agoDone:
Bruce Momjian [Mon, 13 Dec 2004 18:00:51 +0000 (18:00 +0000)]
Done:
<  o Remove per-backend parameter file and move into shared memory

Add for Win32:
>  o Allow the shared memory address to be configured via GUC

19 years agoClearify which version numbers are bumped for major and minor releases.
Bruce Momjian [Mon, 13 Dec 2004 17:28:54 +0000 (17:28 +0000)]
Clearify which version numbers are bumped for major and minor releases.

19 years agoNew translations
Peter Eisentraut [Mon, 13 Dec 2004 16:30:53 +0000 (16:30 +0000)]
New translations

19 years agoTranslation updates
Peter Eisentraut [Mon, 13 Dec 2004 16:28:23 +0000 (16:28 +0000)]
Translation updates

19 years agoAdd:
Bruce Momjian [Mon, 13 Dec 2004 14:21:32 +0000 (14:21 +0000)]
Add:

>
> * Allow the size of the buffer cache used by temporary objects to be
>   specified as a GUC variable
>
>   Larger local buffer cache sizes requires more efficient handling of
>   local cache lookups.

19 years agoUpdate FAQ.
Bruce Momjian [Mon, 13 Dec 2004 03:37:09 +0000 (03:37 +0000)]
Update FAQ.

Euler Taveira de Oliveira

19 years agoAvoid generating excess (and illegal) parentheses around an aliased JOIN
Tom Lane [Mon, 13 Dec 2004 00:33:06 +0000 (00:33 +0000)]
Avoid generating excess (and illegal) parentheses around an aliased JOIN
in prettyprint mode.  Andreas Pflug

19 years agoAdd regression tests for the fix committed by Tom for casting between
Neil Conway [Sun, 12 Dec 2004 22:49:50 +0000 (22:49 +0000)]
Add regression tests for the fix committed by Tom for casting between
the row types of parent/child tables.

19 years agoPREPARE and EXPLAIN need to copy the source query just like we recently
Tom Lane [Sun, 12 Dec 2004 20:17:06 +0000 (20:17 +0000)]
PREPARE and EXPLAIN need to copy the source query just like we recently
had to do in DECLARE CURSOR.  AFAICS these are all the places affected.
PREPARE case per example from Michael Fuhr, EXPLAIN case located by
grepping for planner calls ...

19 years agoAdd missing newlines at end of error messages.
Peter Eisentraut [Sun, 12 Dec 2004 18:26:29 +0000 (18:26 +0000)]
Add missing newlines at end of error messages.

19 years agoTranslation updates
Peter Eisentraut [Sun, 12 Dec 2004 18:09:18 +0000 (18:09 +0000)]
Translation updates

19 years agoFix problems with certain shells (e.g., FreeBSD, Cygwin) clearing the
Peter Eisentraut [Sun, 12 Dec 2004 15:34:15 +0000 (15:34 +0000)]
Fix problems with certain shells (e.g., FreeBSD, Cygwin) clearing the
exit status in multiline traps.

19 years agoUpgrade formrdesc() so that it can correctly initialize the tupledesc
Tom Lane [Sun, 12 Dec 2004 05:07:50 +0000 (05:07 +0000)]
Upgrade formrdesc() so that it can correctly initialize the tupledesc
(rd_att) field of a nailed-in-cache relcache entry.  This fixes the bug
reported by Alvaro 8-Dec-2004; I believe it probably also explains
Grant Finnemore's report of 10-Sep-2004.

In an unrelated change in the same file, put back 7.4's response to
failure to rename() the relcache init file, ie, unlink the useless
temp file.  I did not put back the warning message, since there might
actually be some reason not to have that.

19 years agoInstead of supposing (wrongly, in the general case) that the rowtype
Tom Lane [Sat, 11 Dec 2004 23:26:51 +0000 (23:26 +0000)]
Instead of supposing (wrongly, in the general case) that the rowtype
of an inheritance child table is binary-compatible with the rowtype of
its parent, invent an expression node type that does the conversion
correctly.  Fixes the new bug exhibited by Kris Shannon as well as a
lot of old bugs that would only show up when using multiple inheritance
or after altering the parent table.

19 years agoAdded port reports
Peter Eisentraut [Sat, 11 Dec 2004 20:54:14 +0000 (20:54 +0000)]
Added port reports

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 20:19:24 +0000 (20:19 +0000)]
Translation updates

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 20:10:14 +0000 (20:10 +0000)]
Translation updates

19 years agoPatch to the pl/perl documents that clarifies the scope of global data and
Peter Eisentraut [Sat, 11 Dec 2004 20:03:37 +0000 (20:03 +0000)]
Patch to the pl/perl documents that clarifies the scope of global data and
gives an example of storing a code reference

by David Fetter

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 19:56:07 +0000 (19:56 +0000)]
Translation updates

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 19:45:34 +0000 (19:45 +0000)]
Translation updates

19 years agoAdded port reports
Peter Eisentraut [Sat, 11 Dec 2004 19:31:45 +0000 (19:31 +0000)]
Added port reports

19 years agoAdded port report
Peter Eisentraut [Sat, 11 Dec 2004 19:13:32 +0000 (19:13 +0000)]
Added port report

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 19:09:08 +0000 (19:09 +0000)]
Translation updates

19 years agoTranslation updates
Peter Eisentraut [Sat, 11 Dec 2004 19:03:49 +0000 (19:03 +0000)]
Translation updates

19 years agoAdded port report
Peter Eisentraut [Sat, 11 Dec 2004 19:01:46 +0000 (19:01 +0000)]
Added port report

19 years agoReorder IRC examples.
Bruce Momjian [Fri, 10 Dec 2004 18:01:59 +0000 (18:01 +0000)]
Reorder IRC examples.

19 years agoActiveSnapshot must be set to something valid while running deferred
Tom Lane [Mon, 6 Dec 2004 23:57:17 +0000 (23:57 +0000)]
ActiveSnapshot must be set to something valid while running deferred
triggers during COMMIT.  Per trouble report from Frank van Vugt.

19 years agoStart filling in the supported platforms list with build farm information.
Peter Eisentraut [Mon, 6 Dec 2004 23:47:08 +0000 (23:47 +0000)]
Start filling in the supported platforms list with build farm information.

19 years agoTranslation updates
Peter Eisentraut [Mon, 6 Dec 2004 22:10:42 +0000 (22:10 +0000)]
Translation updates

19 years agoFixed bug in parsing of typedef'ed array sizes.
Michael Meskes [Mon, 6 Dec 2004 20:35:35 +0000 (20:35 +0000)]
Fixed bug in parsing of typedef'ed array sizes.
Synced parser.

19 years agoFix off-by-1 bug in pg_ctl in -D handling, per report from pjmodos@parba.cz
Neil Conway [Mon, 6 Dec 2004 01:09:20 +0000 (01:09 +0000)]
Fix off-by-1 bug in pg_ctl in -D handling, per report from pjmodos@parba.cz

19 years agoFix several mistaken xreflabel entries, including those spotted by
Tom Lane [Sun, 5 Dec 2004 20:05:47 +0000 (20:05 +0000)]
Fix several mistaken xreflabel entries, including those spotted by
Michael Fuhr.

19 years agoBack out addition of Win1252 encoding.
Bruce Momjian [Sat, 4 Dec 2004 18:19:33 +0000 (18:19 +0000)]
Back out addition of Win1252 encoding.

19 years agoProperl format HTML in developer's FAQ.
Bruce Momjian [Sat, 4 Dec 2004 04:12:11 +0000 (04:12 +0000)]
Properl format HTML in developer's FAQ.

19 years agotag configure for rc1 .. REL8_0_0RC1
PostgreSQL Daemon [Fri, 3 Dec 2004 22:24:53 +0000 (22:24 +0000)]
tag configure for rc1 ..

19 years agoAs long as we're forcing an ORDER BY in these four join queries, we had
Tom Lane [Fri, 3 Dec 2004 22:19:28 +0000 (22:19 +0000)]
As long as we're forcing an ORDER BY in these four join queries, we had
better make sure the sort order is totally specified; else we get burnt
by platform-specific behavior of qsort() with equal keys.  Per buildfarm
results.

19 years agoUse StrNCpy not strncpy to fill hash key, to ensure the resulting key
Tom Lane [Fri, 3 Dec 2004 21:26:31 +0000 (21:26 +0000)]
Use StrNCpy not strncpy to fill hash key, to ensure the resulting key
is null-terminated.  I think this is not a real bug because the parser
would always have truncated the identifier to NAMEDATALEN-1 already,
but let's be safe.  Per report from Klocwork.

19 years agoAdjust help message for --no-owner, per Alvaro Herrera.
Tom Lane [Fri, 3 Dec 2004 18:48:19 +0000 (18:48 +0000)]
Adjust help message for --no-owner, per Alvaro Herrera.

19 years agoBack out ARRAY() patch.
Bruce Momjian [Fri, 3 Dec 2004 18:34:31 +0000 (18:34 +0000)]
Back out ARRAY() patch.

19 years agoImprove description of --disable-rpath workaround for regression test
Tom Lane [Fri, 3 Dec 2004 17:46:19 +0000 (17:46 +0000)]
Improve description of --disable-rpath workaround for regression test
problems.

19 years ago> If it bothers you that much. I'd make a flag, cleared at the start of
Bruce Momjian [Fri, 3 Dec 2004 17:13:28 +0000 (17:13 +0000)]
> If it bothers you that much. I'd make a flag, cleared at the start of
> each COPY, and then where we test for CR or LF in CopyAttributeOutCSV,
> if the flag is not set then set it and issue the warning.

Andrew Dunstan

19 years ago>>This patch adds another plpgsql trigger example to the chapter. It uses
Bruce Momjian [Fri, 3 Dec 2004 17:12:09 +0000 (17:12 +0000)]
>>This patch adds another plpgsql trigger example to the chapter. It uses
>>the emp table again, but shows how to audit changes into another table
>>(emp_audit).

Mark Kirkwood

19 years agoAdd:
Bruce Momjian [Fri, 3 Dec 2004 17:07:50 +0000 (17:07 +0000)]
Add:

> * ANALYZE should record a pg_statistic entry for an all-NULL column

19 years agoTypo fixes.
Bruce Momjian [Fri, 3 Dec 2004 06:30:40 +0000 (06:30 +0000)]
Typo fixes.

19 years agoFix typo.
Bruce Momjian [Fri, 3 Dec 2004 06:18:12 +0000 (06:18 +0000)]
Fix typo.

19 years agoMerge pgadmin paragraph.
Bruce Momjian [Fri, 3 Dec 2004 06:04:38 +0000 (06:04 +0000)]
Merge pgadmin paragraph.

19 years agoFirst version of external projects manual section.
Bruce Momjian [Fri, 3 Dec 2004 05:50:18 +0000 (05:50 +0000)]
First version of external projects manual section.

19 years agoAdd:
Bruce Momjian [Fri, 3 Dec 2004 02:24:01 +0000 (02:24 +0000)]
Add:

> * Fix priority ordering of read and write light-weight locks (Neil)

19 years agoFix compile breakage from SIGPIPE fix for threading.
Bruce Momjian [Fri, 3 Dec 2004 01:58:43 +0000 (01:58 +0000)]
Fix compile breakage from SIGPIPE fix for threading.

19 years agoRemove doc changes that were in process during previous commit.
Bruce Momjian [Fri, 3 Dec 2004 01:23:38 +0000 (01:23 +0000)]
Remove doc changes that were in process during previous commit.

19 years agoBack out fix for Unicode characters above 0x10000
Bruce Momjian [Fri, 3 Dec 2004 01:20:33 +0000 (01:20 +0000)]
Back out fix for Unicode characters above 0x10000

19 years agoCode review for recent libpq changes. Be more careful about error
Tom Lane [Thu, 2 Dec 2004 23:20:21 +0000 (23:20 +0000)]
Code review for recent libpq changes.  Be more careful about error
handling in SIGPIPE processing; avoid unnecessary pollution of application
link-symbol namespace; spell 'pointer to function' in the conventional
way.

19 years agoAdd:
Bruce Momjian [Thu, 2 Dec 2004 23:11:30 +0000 (23:11 +0000)]
Add:

<
>  o Fix problem with shared memory on the Win32 Terminal Server

19 years ago> I have installed your patch and adjusted the names of the standards
Bruce Momjian [Thu, 2 Dec 2004 22:51:28 +0000 (22:51 +0000)]
> I have installed your patch and adjusted the names of the standards
> throughout to the spellings suggested by your book.

Great.

A follow-up patch for current CVS HEAD is attached, and available at
http://troels.arvin.dk/db/pgsql/conformance/pgsql-sql-conformance-
followup.patch

The patch
 - includes a core feature ID that had been left
   out by mistake (C011)
 - updates the sql_feature_packages.txt table to
   reflect changes in SQL:2003 which were not
   covered properly in my last patch

Troels Arvin

19 years ago> > 8.0beta3 has pg_autovacuum included, when I want to run this as a
Bruce Momjian [Thu, 2 Dec 2004 22:48:10 +0000 (22:48 +0000)]
> > 8.0beta3 has pg_autovacuum included, when I want to run this as a
> > Windows service, it says you can use the -I and -R options.
> >
> > When I do that and I specify a password with '-P'
> (uppercase) then in
> > the registry it's saved as '-p' (lowercase) in the
> service-commandline
> > (ImagePath).

This was fixed in v1.21 of pg_autovacuum.c, That rev is tagged for
beta3, so you should not be seeing this issue unless you actually have
an older version for some reason.

http://developer.postgresql.org/cvsweb.cgi/pgsql/contrib/pg_autovacuum/p
g_autovacuum.c.diff?r1=1.20;r2=1.21;f=h

> > Also it removes the quotes I added and I'm not so sure it
> would work
> > the way it's supposed to, without it.

It's not so much that it strips them (that happens automagically), more
that it doesn't re-add them when it writes the command line in the
registry. The attached patch fixes that by simply quoting all options
that may need it.
> > If you add DependOnService (a REG_MULTI_SZ an
> array-like-thingie) and
> > have the name (in this case: pgsql-8.0-beta2-dev3) of a service it
> > depends on, it will not fail to start (it will not even try, as
> > PostgreSQL is not running), when PostgreSQL already failed.
> >
> > Maybe it's an idea to specify it on the commandline (what
> service to
> > depend on).

A -E <service> option is added in the attached patch.

Dave Page

19 years ago> I enclose a short patch to reduce the PGARCH_RESTART_INTERVAL from 60
Bruce Momjian [Thu, 2 Dec 2004 22:40:18 +0000 (22:40 +0000)]
> I enclose a short patch to reduce the PGARCH_RESTART_INTERVAL from 60
> seconds to 10 seconds. The original number was plucked from thin air
> some months ago, and I'd like to review that now based upon further
> thought, observation and experience.
>
> This change has little or no effect on performance, since the interval
> is there mainly to avoid repeated respawn attempts if archiver fails at
> startup. Archiver start-up time is very quick, so there is little danger
> of exceeding 10 seconds.
>
> On a busy system, if the archiver does die, then many files can build up
> in the 60 seconds before respawning. That xlog file backlog could take
> some time to clear. This then leaves a larger than normal window of data
> loss for a possibly long period.
>
> It's a minor change only, with no other effect on function.

Simon Riggs