]> granicus.if.org Git - postgresql/log
postgresql
23 years agoAdd multibyte support
Hiroshi Inoue [Mon, 26 Mar 2001 09:34:17 +0000 (09:34 +0000)]
Add multibyte support

23 years agoFix typo
Tatsuo Ishii [Mon, 26 Mar 2001 08:32:10 +0000 (08:32 +0000)]
Fix typo

23 years agoFix unportable assumptions about alignment of local char[n] variables.
Tom Lane [Sun, 25 Mar 2001 23:23:59 +0000 (23:23 +0000)]
Fix unportable assumptions about alignment of local char[n] variables.

23 years agoImprove comments for xlog item size #defines.
Tom Lane [Sun, 25 Mar 2001 22:40:58 +0000 (22:40 +0000)]
Improve comments for xlog item size #defines.

23 years agoUpdate the changes for version 3.2.
D'Arcy J.M. Cain [Sun, 25 Mar 2001 22:23:44 +0000 (22:23 +0000)]
Update the changes for version 3.2.

23 years agoDocument the --with-python flag as a simpler way of installing the
D'Arcy J.M. Cain [Sun, 25 Mar 2001 22:22:51 +0000 (22:22 +0000)]
Document the --with-python flag as a simpler way of installing the
PyGreSQL module when installing PostgreSQL.

Document the location of the WIN32 binaries.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 25 Mar 2001 21:41:52 +0000 (21:41 +0000)]
Update TODO list.

23 years agoPick up any extra -I options for Python build.
Peter Eisentraut [Sun, 25 Mar 2001 19:44:04 +0000 (19:44 +0000)]
Pick up any extra -I options for Python build.

23 years agoFix unportable assumptions about alignment of local char[n] variables.
Tom Lane [Sun, 25 Mar 2001 19:30:28 +0000 (19:30 +0000)]
Fix unportable assumptions about alignment of local char[n] variables.

23 years agoThe regexp pattern characters ^ and $ should be explained as matching
Tom Lane [Sun, 25 Mar 2001 18:14:31 +0000 (18:14 +0000)]
The regexp pattern characters ^ and $ should be explained as matching
at the beginning and end of the input string, not the beginning and end
of "a line", since Postgres does not allow them to match at newline
characters in the data.

23 years agoRe-order declarations to un-break the non-HAS_TEST_AND_SET case.
Tom Lane [Sun, 25 Mar 2001 17:52:46 +0000 (17:52 +0000)]
Re-order declarations to un-break the non-HAS_TEST_AND_SET case.

23 years agospell fix
Peter Eisentraut [Sun, 25 Mar 2001 10:16:12 +0000 (10:16 +0000)]
spell fix

23 years agoAllow building documentation outside source tree.
Peter Eisentraut [Sun, 25 Mar 2001 08:32:24 +0000 (08:32 +0000)]
Allow building documentation outside source tree.
from Ian Lance Taylor

23 years agoFix code that incorrectly assumed a 'char foo[N]' local variable would
Tom Lane [Sun, 25 Mar 2001 00:45:20 +0000 (00:45 +0000)]
Fix code that incorrectly assumed a 'char foo[N]' local variable would
be aligned on a word boundary.  Per report from Steve Nicolai.

23 years agoMake less misleading message for failed + ignored tests.
Peter Eisentraut [Sat, 24 Mar 2001 23:32:25 +0000 (23:32 +0000)]
Make less misleading message for failed + ignored tests.

23 years agoPortability fix from Steve Nicolai.
Tom Lane [Sat, 24 Mar 2001 23:11:14 +0000 (23:11 +0000)]
Portability fix from Steve Nicolai.

23 years agominor copy editing
Peter Eisentraut [Sat, 24 Mar 2001 23:03:26 +0000 (23:03 +0000)]
minor copy editing

23 years agoDisable pg_upgrade.
Peter Eisentraut [Sat, 24 Mar 2001 13:21:14 +0000 (13:21 +0000)]
Disable pg_upgrade.

23 years agoHadn't built on a PARISC 1.1 box in a long time ... but when I did,
Tom Lane [Sat, 24 Mar 2001 06:17:41 +0000 (06:17 +0000)]
Hadn't built on a PARISC 1.1 box in a long time ... but when I did,
it turned out that the libpq makefile chokes on '-L /lib/pa1.1'.
Need to remove the space so filter rule works.

23 years agoUpdate info about mailing lists, make a few other minor improvements.
Tom Lane [Sat, 24 Mar 2001 03:40:44 +0000 (03:40 +0000)]
Update info about mailing lists, make a few other minor improvements.

23 years agoUpdate to the current state of platform support.
Thomas G. Lockhart [Sat, 24 Mar 2001 02:36:01 +0000 (02:36 +0000)]
Update to the current state of platform support.
Still need some tested for the 7.1 release.

23 years agoFix up wording slightly. No change in content.
Thomas G. Lockhart [Sat, 24 Mar 2001 02:35:25 +0000 (02:35 +0000)]
Fix up wording slightly. No change in content.

23 years agoupdate the ChangeLog in preparation for release ...
Marc G. Fournier [Sat, 24 Mar 2001 00:48:17 +0000 (00:48 +0000)]
update the ChangeLog in preparation for release ...

23 years agoMiscellaneous updates and minor copy-editing.
Tom Lane [Sat, 24 Mar 2001 00:24:26 +0000 (00:24 +0000)]
Miscellaneous updates and minor copy-editing.

23 years agoDoc update from Roberto Mello: improved versions of instr() examples.
Tom Lane [Fri, 23 Mar 2001 22:07:50 +0000 (22:07 +0000)]
Doc update from Roberto Mello: improved versions of instr() examples.

23 years agoWhen using 'long long int' for int64 type, check to see if the compiler
Tom Lane [Fri, 23 Mar 2001 18:42:12 +0000 (18:42 +0000)]
When using 'long long int' for int64 type, check to see if the compiler
accepts nnnLL syntax for long long constants.  If so, decorate the CRC64
constants with LL to avoid warnings and/or erroneous results from certain
non-standards-compliant compilers.

23 years agoMark exception and assert global variables as DLLIMPORT, so that plpgsql
Tom Lane [Fri, 23 Mar 2001 18:26:01 +0000 (18:26 +0000)]
Mark exception and assert global variables as DLLIMPORT, so that plpgsql
can be compiled with asserts enabled on Windoze.

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 23 Mar 2001 17:09:09 +0000 (17:09 +0000)]
Update TODO list.

23 years agoAdd checking for strtoul. SunOS4 does not have it.
Tatsuo Ishii [Fri, 23 Mar 2001 05:46:05 +0000 (05:46 +0000)]
Add checking for strtoul. SunOS4 does not have it.

23 years agoFix comments that were mis-wrapped, for Tom Lane.
Bruce Momjian [Fri, 23 Mar 2001 04:49:58 +0000 (04:49 +0000)]
Fix comments that were mis-wrapped, for Tom Lane.

23 years agoFix SPI example to reflect new-style calling convention for textout().
Tom Lane [Fri, 23 Mar 2001 01:51:08 +0000 (01:51 +0000)]
Fix SPI example to reflect new-style calling convention for textout().

23 years agoFix checking of file operations in demo ('files') output format.
Philip Warner [Fri, 23 Mar 2001 01:27:12 +0000 (01:27 +0000)]
Fix checking of file operations in demo ('files') output format.

23 years agoFix problems with coredumps due to ^C when longjmp buffer isn't valid.
Tom Lane [Fri, 23 Mar 2001 00:36:38 +0000 (00:36 +0000)]
Fix problems with coredumps due to ^C when longjmp buffer isn't valid.
Now, we will only catch ^C at times when it is valid.

23 years agoFix Joubert's complaint that int8-sized numeric literals are mishandled
Tom Lane [Thu, 22 Mar 2001 17:41:47 +0000 (17:41 +0000)]
Fix Joubert's complaint that int8-sized numeric literals are mishandled
on Alpha (because parser mistakenly assumes that a nonoverflow result
from strtol means the value will fit into int4).  A scan for other uses
of strtol and strtoul found a couple other places with the same mistake;
fix them too.  The changes are all conditional on HAVE_LONG_INT_64 to
avoid complaints from compilers that think x != x is a silly test
(cf. pg_atoi).

23 years agoRun pgindent on _int.c, for Tom Lane. I had skipped this file because
Bruce Momjian [Thu, 22 Mar 2001 15:24:12 +0000 (15:24 +0000)]
Run pgindent on _int.c, for Tom Lane.  I had skipped this file because
of a complex macro that pgindent complained about.  Hand-adjusted macro
for pgindent run.

23 years agogeometry tests 'FAILED' from a precision standpoint, but differently then
Marc G. Fournier [Thu, 22 Mar 2001 15:13:19 +0000 (15:13 +0000)]
geometry tests 'FAILED' from a precision standpoint, but differently then
what solaris-precision takes into account ... its possible that solaris-precision
is 'stale', but created a new one "just in case" ...

23 years agoRemove dashes in comments that don't need them, rewrap with pgindent.
Bruce Momjian [Thu, 22 Mar 2001 06:16:21 +0000 (06:16 +0000)]
Remove dashes in comments that don't need them, rewrap with pgindent.

23 years agopgindent run. Make it all clean.
Bruce Momjian [Thu, 22 Mar 2001 04:01:46 +0000 (04:01 +0000)]
pgindent run.  Make it all clean.

23 years agocorrection
Peter Eisentraut [Wed, 21 Mar 2001 19:09:03 +0000 (19:09 +0000)]
correction

23 years agoThe attached patch fixes win32.mak which got broken by the Multibyte mod
Hiroshi Inoue [Wed, 21 Mar 2001 16:26:17 +0000 (16:26 +0000)]
The attached patch fixes win32.mak which got broken by the Multibyte mod
:-(, and tidies it up a little as well.

Regards, Dave.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 21 Mar 2001 15:46:11 +0000 (15:46 +0000)]
Update TODO list.

23 years agoCheck bufHdr->cntxDirty and call StartBufferIO in BufferSync()
Vadim B. Mikheev [Wed, 21 Mar 2001 10:13:29 +0000 (10:13 +0000)]
Check bufHdr->cntxDirty and call StartBufferIO in BufferSync()
*before* acquiring shlock on buffer context. This way we should be
protected against conflicts with FlushRelationBuffers.
(Seems we never do excl lock and then StartBufferIO for the same
buffer, so there should be no deadlock here, - but we'd better
check this very soon).

23 years agoa fix from Eiji Tokuya.
Hiroshi Inoue [Wed, 21 Mar 2001 09:35:25 +0000 (09:35 +0000)]
a fix from Eiji Tokuya.

23 years agoAdd mmap info. Seems mmap may not be a good idea.
Bruce Momjian [Wed, 21 Mar 2001 04:39:28 +0000 (04:39 +0000)]
Add mmap info.  Seems mmap may not be a good idea.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 21 Mar 2001 04:30:26 +0000 (04:30 +0000)]
Update TODO list.

23 years agowell, here goes our first Release Candidate for 7.1 *cross fingers*
Marc G. Fournier [Tue, 20 Mar 2001 23:35:20 +0000 (23:35 +0000)]
well, here goes our first Release Candidate for 7.1 *cross fingers*

23 years agoFix 'ps' WAIT status display bug on setproctitle() platforms, cleared by
Bruce Momjian [Tue, 20 Mar 2001 22:31:54 +0000 (22:31 +0000)]
Fix 'ps' WAIT status display bug on setproctitle() platforms, cleared by
Tom Lane.

23 years agoUpdate Japanese FAQ.
Bruce Momjian [Tue, 20 Mar 2001 22:15:43 +0000 (22:15 +0000)]
Update Japanese FAQ.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 20 Mar 2001 22:04:59 +0000 (22:04 +0000)]
Update TODO list.

23 years agoRecommend CREATE TABLE AS in preference to SELECT INTO. Remove the
Tom Lane [Tue, 20 Mar 2001 20:54:41 +0000 (20:54 +0000)]
Recommend CREATE TABLE AS in preference to SELECT INTO.  Remove the
(inadequate anyway) mention of SELECT INTO from the main SELECT ref page.
Point out that SELECT INTO means something else in plpgsql and ecpg.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 20 Mar 2001 20:32:30 +0000 (20:32 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 20 Mar 2001 20:31:56 +0000 (20:31 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 20 Mar 2001 20:31:07 +0000 (20:31 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 20 Mar 2001 19:56:25 +0000 (19:56 +0000)]
Update TODO list.

23 years agoRepair intarray's problems with TOASTed input, avoid scribbling on
Tom Lane [Tue, 20 Mar 2001 03:08:12 +0000 (03:08 +0000)]
Repair intarray's problems with TOASTed input, avoid scribbling on
source data, improve split algorithm for intbig_ops.  Oleg Bartunov
with some kibitzing from Tom Lane.

23 years agoMention that parallel regress script needs to be run under ksh not sh
Tom Lane [Tue, 20 Mar 2001 00:18:21 +0000 (00:18 +0000)]
Mention that parallel regress script needs to be run under ksh not sh
on HPUX.

23 years agoDiscuss LOCALE differences as a reason for regression test failure.
Tom Lane [Tue, 20 Mar 2001 00:09:36 +0000 (00:09 +0000)]
Discuss LOCALE differences as a reason for regression test failure.

23 years agoTweak out-of-memory error messages to include the request size, so that
Tom Lane [Mon, 19 Mar 2001 22:29:39 +0000 (22:29 +0000)]
Tweak out-of-memory error messages to include the request size, so that
it's easier to tell whether a bug report is talking about progressive
memory exhaustion or a wacko requested chunk size.

23 years agoMake regular-expression error messages a tad less obscure,
Tom Lane [Mon, 19 Mar 2001 22:27:46 +0000 (22:27 +0000)]
Make regular-expression error messages a tad less obscure,
per gripe from Josh Berkus.

23 years agoUpdate Peter Mount's email address in README.
Bruce Momjian [Mon, 19 Mar 2001 21:57:09 +0000 (21:57 +0000)]
Update Peter Mount's email address in README.

23 years agoAdd missing TO.
Peter Eisentraut [Mon, 19 Mar 2001 16:20:52 +0000 (16:20 +0000)]
Add missing TO.

23 years agoAdd note about saving large objects during upgrade.
Peter Eisentraut [Mon, 19 Mar 2001 16:19:26 +0000 (16:19 +0000)]
Add note about saving large objects during upgrade.

23 years ago - Added CVS headers to files
Philip Warner [Mon, 19 Mar 2001 02:35:29 +0000 (02:35 +0000)]
 - Added CVS headers to files
 - Avoid forcing table name to lower case in FixupBlobXrefs
 - Removed fmtId calls for all ArchiveEntry name fields. This fixes
   quoting problems in trigger enable/disable code for mixed case
   table names, and avoids commands like 'pg_restore -t '"TblA"'

23 years agoRestore include-file updates, and fix some of the more glaring macro
Tom Lane [Sun, 18 Mar 2001 23:36:55 +0000 (23:36 +0000)]
Restore include-file updates, and fix some of the more glaring macro
sloppiness (insufficient parenthesization, etc).
It still fails regress test for me, however.

23 years agoRestore Peter's makefile fixes.
Tom Lane [Sun, 18 Mar 2001 23:35:18 +0000 (23:35 +0000)]
Restore Peter's makefile fixes.

23 years agoGive postmaster enough time to update the pidfile before checking to
Tom Lane [Sun, 18 Mar 2001 20:27:11 +0000 (20:27 +0000)]
Give postmaster enough time to update the pidfile before checking to
see if it's been updated.  Skip the whole mess if we didn't see any
old pidfile.

23 years agoRemove NEXTXID xlog record type to avoid three-way deadlock risk.
Tom Lane [Sun, 18 Mar 2001 20:18:59 +0000 (20:18 +0000)]
Remove NEXTXID xlog record type to avoid three-way deadlock risk.
NEXTXID isn't really necessary, per previous discussion in pghackers,
but I mulishy insisted we should put it in anyway.  Mea culpa.

23 years agoWhen we add 'waiting' to the ps_status display, there should be a
Tom Lane [Sun, 18 Mar 2001 20:13:13 +0000 (20:13 +0000)]
When we add 'waiting' to the ps_status display, there should be a
space in front of it.  Improve comments a little.

23 years agoConsidering the BeOS port didn't compile without IPC_STAT and shm_nattch,
Tom Lane [Sun, 18 Mar 2001 18:32:02 +0000 (18:32 +0000)]
Considering the BeOS port didn't compile without IPC_STAT and shm_nattch,
I'm betting the QNX4 port does not either ...

23 years agoPatches from Cyril Velter to make shared-memory-conflict-detection code
Tom Lane [Sun, 18 Mar 2001 18:22:08 +0000 (18:22 +0000)]
Patches from Cyril Velter to make shared-memory-conflict-detection code
work in BeOS port.

23 years agoFixed but in _enableTriggersIfNecessary introduced by previous patches by me
Philip Warner [Sun, 18 Mar 2001 03:45:12 +0000 (03:45 +0000)]
Fixed but in _enableTriggersIfNecessary introduced by previous patches by me
which caused segfault in a data-only retore with BLOBs

23 years agoFix coredump due to writing one more byte than we'd allocated.
Tom Lane [Sun, 18 Mar 2001 02:25:32 +0000 (02:25 +0000)]
Fix coredump due to writing one more byte than we'd allocated.

23 years agoMake sure ControlFile logId/logSeg don't go backwards (barely possible given a
Tom Lane [Sun, 18 Mar 2001 00:30:27 +0000 (00:30 +0000)]
Make sure ControlFile logId/logSeg don't go backwards (barely possible given a
slow backend, if we update unconditionally as the code did before).

23 years agoUpdate contrib intarray to Jan 25 version.
Bruce Momjian [Sat, 17 Mar 2001 21:59:42 +0000 (21:59 +0000)]
Update contrib intarray to Jan 25 version.

23 years agoRearrange XLogFileInit so that control-file spinlock is not held while filling
Tom Lane [Sat, 17 Mar 2001 20:54:13 +0000 (20:54 +0000)]
Rearrange XLogFileInit so that control-file spinlock is not held while filling
the new log file with zeroes, only while renaming it into place.  This should
prevent problems with 'stuck spinlock' errors under heavy load.

23 years agoReformat the "major changes" to 72 or so columns.
Peter Eisentraut [Sat, 17 Mar 2001 18:15:36 +0000 (18:15 +0000)]
Reformat the "major changes" to 72 or so columns.

23 years agoRepair.
Peter Eisentraut [Sat, 17 Mar 2001 18:08:14 +0000 (18:08 +0000)]
Repair.

23 years agoTuning for docbook2man.
Peter Eisentraut [Sat, 17 Mar 2001 16:27:31 +0000 (16:27 +0000)]
Tuning for docbook2man.

23 years agoNew updates from Roberto Mello sent privately today due to email troubles.
Thomas G. Lockhart [Sat, 17 Mar 2001 01:53:22 +0000 (01:53 +0000)]
New updates from Roberto Mello sent privately today due to email troubles.

23 years agoMake note that version 1.1.8 doesn't work.
Peter Eisentraut [Fri, 16 Mar 2001 22:12:13 +0000 (22:12 +0000)]
Make note that version 1.1.8 doesn't work.

23 years agoGenerated Changelog since last Beta ...
Marc G. Fournier [Fri, 16 Mar 2001 21:49:51 +0000 (21:49 +0000)]
Generated Changelog since last Beta ...

23 years agoupgrade version to beta6
Marc G. Fournier [Fri, 16 Mar 2001 21:29:15 +0000 (21:29 +0000)]
upgrade version to beta6

23 years agoDocumentation update from Marko Kreen.
Tom Lane [Fri, 16 Mar 2001 17:42:56 +0000 (17:42 +0000)]
Documentation update from Marko Kreen.

23 years agoFixed bug in handling of pointers to structs.
Michael Meskes [Fri, 16 Mar 2001 13:05:57 +0000 (13:05 +0000)]
Fixed bug in handling of pointers to structs.

23 years agoOops I forgot to add new files for multibyte support.
Hiroshi Inoue [Fri, 16 Mar 2001 13:03:04 +0000 (13:03 +0000)]
Oops I forgot to add new files for multibyte support.
Sorry Eiji.

23 years agoSupport syncing WAL log to disk using either fsync(), fdatasync(),
Tom Lane [Fri, 16 Mar 2001 05:44:33 +0000 (05:44 +0000)]
Support syncing WAL log to disk using either fsync(), fdatasync(),
O_SYNC, or O_DSYNC (as available on a given platform).  Add GUC parameter
to control sync method.
Also, add defense to XLogWrite to prevent it from going nuts if passed
a target write position that's past the end of the buffers so far filled
by XLogInsert.

23 years agoResetting the XLOG can't include rewinding it to logical position zero,
Tom Lane [Fri, 16 Mar 2001 05:08:39 +0000 (05:08 +0000)]
Resetting the XLOG can't include rewinding it to logical position zero,
because we need page LSNs stored in the main database to be less than
the current XLOG position.  Hence, generate the new XLOG segment at last
old segment number plus one.

23 years agoAdd multibyte support.
Hiroshi Inoue [Fri, 16 Mar 2001 01:17:23 +0000 (01:17 +0000)]
Add multibyte support.
Provide an extenisible scheme of encoding conversion.
As the first step, SJIS and BIG5 are supported.
From now on multibyte people would be happy to use
this psqlodbc driver.

Eiji Tokuya e-tokuya@mail.sankyo-unyu.co.jp

23 years agoUpdate sample error messages to agree with current code.
Tom Lane [Thu, 15 Mar 2001 20:01:32 +0000 (20:01 +0000)]
Update sample error messages to agree with current code.

23 years agoFix parameter handling.
D'Arcy J.M. Cain [Thu, 15 Mar 2001 11:56:05 +0000 (11:56 +0000)]
Fix parameter handling.

Fix a bug where cs.execute('select %d + %d', (1, 2)) would get interpreted
as cs.executemany('select %d + %d', (1, 2))

23 years agoAdd WIN32 support.
D'Arcy J.M. Cain [Thu, 15 Mar 2001 11:26:22 +0000 (11:26 +0000)]
Add WIN32 support.

23 years agoImprove description of to_char templates.
Tom Lane [Thu, 15 Mar 2001 01:07:51 +0000 (01:07 +0000)]
Improve description of to_char templates.

23 years agoGive a more reasonable error message for a bad attribute name applied
Tom Lane [Wed, 14 Mar 2001 23:55:33 +0000 (23:55 +0000)]
Give a more reasonable error message for a bad attribute name applied
to a join or subselect alias ... cf. Oliver Elphick's complaint 13-Mar.

23 years agoFix markup.
Peter Eisentraut [Wed, 14 Mar 2001 22:20:02 +0000 (22:20 +0000)]
Fix markup.

23 years agoReduce amount of memory used per tuple for after-event triggers. This
Tom Lane [Wed, 14 Mar 2001 21:50:32 +0000 (21:50 +0000)]
Reduce amount of memory used per tuple for after-event triggers.  This
is still a memory leak, but a little less bad than it was.

23 years agoCOPY should handle after-insert triggers the same as execMain.c does.
Tom Lane [Wed, 14 Mar 2001 21:47:50 +0000 (21:47 +0000)]
COPY should handle after-insert triggers the same as execMain.c does.
I'm not sure that it's really necessary to save insert events when there
are only after update or delete triggers, but certainly it's wrong for
COPY to behave differently from an INSERT query.

23 years agoA much better version of this is now in backend/port/aix.
Peter Eisentraut [Wed, 14 Mar 2001 21:16:57 +0000 (21:16 +0000)]
A much better version of this is now in backend/port/aix.

23 years agoIt's "gmake dist" now.
Peter Eisentraut [Wed, 14 Mar 2001 21:11:53 +0000 (21:11 +0000)]
It's "gmake dist" now.

23 years agoFor some ungodly reason my etags doesn't have an --output option, but it
Peter Eisentraut [Wed, 14 Mar 2001 21:07:16 +0000 (21:07 +0000)]
For some ungodly reason my etags doesn't have an --output option, but it
does have -o.

23 years agoWe're past Postgres95 now, and I'm sure Jolly doesn't want to receive any
Peter Eisentraut [Wed, 14 Mar 2001 21:03:38 +0000 (21:03 +0000)]
We're past Postgres95 now, and I'm sure Jolly doesn't want to receive any
bug reports.  ;-)