]> granicus.if.org Git - postgresql/log
postgresql
18 years agoSpellchecking run, final cleanups
Peter Eisentraut [Fri, 4 Nov 2005 23:14:02 +0000 (23:14 +0000)]
Spellchecking run, final cleanups

18 years agoUpdate release notes for 8.1.
Bruce Momjian [Fri, 4 Nov 2005 22:21:33 +0000 (22:21 +0000)]
Update release notes for 8.1.

18 years agoRemove a gratuitous string difference (does not affect translations).
Peter Eisentraut [Fri, 4 Nov 2005 22:19:04 +0000 (22:19 +0000)]
Remove a gratuitous string difference (does not affect translations).

18 years agoFix logical error in option description.
Peter Eisentraut [Fri, 4 Nov 2005 22:18:26 +0000 (22:18 +0000)]
Fix logical error in option description.

18 years agoEnsure that we only create one ConsoleCtrlHandler per psql process,
Tom Lane [Fri, 4 Nov 2005 18:35:40 +0000 (18:35 +0000)]
Ensure that we only create one ConsoleCtrlHandler per psql process,
so as to avoid performance issues and possible ultimate crash on long
psql scripts.  Per Merlin Moncure.

18 years agoWe haven't seen any agreement emerge as to what is causing AIX 5.3 ML3
Bruce Momjian [Fri, 4 Nov 2005 18:16:50 +0000 (18:16 +0000)]
We haven't seen any agreement emerge as to what is causing AIX 5.3 ML3
to fail to successfully build the release candidates.

However, a patch has emerged (thanks, Seneca!) that does allow it to
work, and which I'd expect to be portable (better still!).

We are still actively pursuing why it breaks, but supposing that still
remains outstanding, at least the following would allow AIX users to
better survive a build...

Chris Browne

18 years agoDisregard superuserness when checking to see if a role GRANT would
Tom Lane [Fri, 4 Nov 2005 17:25:15 +0000 (17:25 +0000)]
Disregard superuserness when checking to see if a role GRANT would
create circularity of role memberships.  This is a minimum-impact fix
for the problem reported by Florian Pflug.  I thought about removing
the superuser_arg test from is_member_of_role() altogether, as it seems
redundant for many of the callers --- but not all, and it's way too late
in the 8.1 cycle to be making large changes.  Perhaps reconsider this
later.

18 years agoUpdate supported-platforms list.
Tom Lane [Fri, 4 Nov 2005 16:15:20 +0000 (16:15 +0000)]
Update supported-platforms list.

19 years agoAvoid referring to a specific version of the SQL standard except where
Tom Lane [Fri, 4 Nov 2005 02:56:31 +0000 (02:56 +0000)]
Avoid referring to a specific version of the SQL standard except where
necessary, and be careful to refer to the right version where it is
useful to do so.  This partially reverts an ill-considered search and
replace from a few months ago.

19 years agoFix a couple of missed None -> DestNone in comments.
Tom Lane [Thu, 3 Nov 2005 21:35:57 +0000 (21:35 +0000)]
Fix a couple of missed None -> DestNone in comments.

19 years agoFix one overlooked ocurrence of "None" in EXEC_BACKEND block.
Alvaro Herrera [Thu, 3 Nov 2005 20:02:50 +0000 (20:02 +0000)]
Fix one overlooked ocurrence of "None" in EXEC_BACKEND block.

19 years agoFix incorrect header size macros
Teodor Sigaev [Thu, 3 Nov 2005 18:16:31 +0000 (18:16 +0000)]
Fix incorrect header size macros

19 years agoThinking further, it seems we had better also copy down resorigtbl/resorigcol
Tom Lane [Thu, 3 Nov 2005 17:45:29 +0000 (17:45 +0000)]
Thinking further, it seems we had better also copy down resorigtbl/resorigcol
to ensure that SubqueryScan elimination doesn't change the behavior of
reporting of original column sources.

19 years agoFix the recently-added code that eliminates unnecessary SubqueryScan nodes
Tom Lane [Thu, 3 Nov 2005 17:34:03 +0000 (17:34 +0000)]
Fix the recently-added code that eliminates unnecessary SubqueryScan nodes
from a finished plan tree.  We have to copy the output column names
(resname fields) from the SubqueryScan down to its child plan node;
else, if this is the topmost level of the plan, the wrong column names
will be delivered to the client.  Per bug #2017 reported by Jolly Chen.

19 years agoRename the members of CommandDest enum so they don't collide with other uses of
Alvaro Herrera [Thu, 3 Nov 2005 17:11:40 +0000 (17:11 +0000)]
Rename the members of CommandDest enum so they don't collide with other uses of
those names.  (Debug and None were pretty bad names anyway.)  I hope I catched
all uses of the names in comments too.

19 years agoTranslation updates.
Alvaro Herrera [Thu, 3 Nov 2005 17:00:54 +0000 (17:00 +0000)]
Translation updates.

19 years agopgcrypto documentation polishing from Marko Kreen, and a small amount
Tom Lane [Thu, 3 Nov 2005 02:54:07 +0000 (02:54 +0000)]
pgcrypto documentation polishing from Marko Kreen, and a small amount
of copy-editing from myself.

19 years agoSome minor improvements to the CE docs. Also fix a bit of SGML markup
Neil Conway [Thu, 3 Nov 2005 00:51:43 +0000 (00:51 +0000)]
Some minor improvements to the CE docs. Also fix a bit of SGML markup
elsewhere.

19 years agoFix longstanding race condition in transaction log management: there was a
Tom Lane [Thu, 3 Nov 2005 00:23:36 +0000 (00:23 +0000)]
Fix longstanding race condition in transaction log management: there was a
very narrow window in which SimpleLruReadPage or SimpleLruWritePage could
think that I/O was needed when it wasn't (and indeed the buffer had already
been assigned to another page).  This would result in an Assert failure if
Asserts were enabled, and probably in silent data corruption if not.
Reported independently by Jim Nasby and Robert Creager.

I intend a more extensive fix when 8.2 development starts, but this is a
reasonably low-impact patch for the existing branches.

19 years agoAdd Tomoaki Sato to pgbench changes in release notes.
Bruce Momjian [Wed, 2 Nov 2005 14:08:43 +0000 (14:08 +0000)]
Add Tomoaki Sato to pgbench changes in release notes.

19 years agoAdd some documentation for constraint exclusion and basic partitioning.
Neil Conway [Tue, 1 Nov 2005 23:19:05 +0000 (23:19 +0000)]
Add some documentation for constraint exclusion and basic partitioning.
From Simon Riggs; cleanup and editorialization by Neil Conway.

19 years agoMake an editorial pass over the reference pages.
Tom Lane [Tue, 1 Nov 2005 21:09:51 +0000 (21:09 +0000)]
Make an editorial pass over the reference pages.

19 years agoProvide a --no-locale option for pg_regress and a corresponding NOLOCALE=1
Andrew Dunstan [Tue, 1 Nov 2005 15:09:11 +0000 (15:09 +0000)]
Provide a --no-locale option for pg_regress and a corresponding NOLOCALE=1
setting for the regression makefile, allowing Windows users to force locale
settings since Windows does not get its locale from the environment.

Per Petr Jelinek.

19 years agoUpdate supported-platforms entry for Tru64.
Tom Lane [Mon, 31 Oct 2005 21:12:35 +0000 (21:12 +0000)]
Update supported-platforms entry for Tru64.

19 years agoMention in the docs that temporary/non-temp tables can not share
Bruce Momjian [Mon, 31 Oct 2005 18:13:52 +0000 (18:13 +0000)]
Mention in the docs that temporary/non-temp tables can not share
referential integrity relationships.

19 years agoAdd sanity check of query
Teodor Sigaev [Mon, 31 Oct 2005 13:47:09 +0000 (13:47 +0000)]
Add sanity check of query

19 years agoUpdate supported-platforms list.
Tom Lane [Sun, 30 Oct 2005 19:21:42 +0000 (19:21 +0000)]
Update supported-platforms list.

19 years agotag it for rc1 REL8_1_0RC1
PostgreSQL Daemon [Sun, 30 Oct 2005 05:06:03 +0000 (05:06 +0000)]
tag it for rc1

19 years agoFix psql ref typo.
Bruce Momjian [Sun, 30 Oct 2005 03:01:49 +0000 (03:01 +0000)]
Fix psql ref typo.

Andreas Seltenreich

19 years agoFix some allocation size calculation errors that would cause pgbench
Tom Lane [Sat, 29 Oct 2005 19:38:07 +0000 (19:38 +0000)]
Fix some allocation size calculation errors that would cause pgbench
to fail with large test scripts.

19 years agoUpdate a couple of obsolete comments.
Tom Lane [Sat, 29 Oct 2005 18:39:17 +0000 (18:39 +0000)]
Update a couple of obsolete comments.

19 years agoTranslation updates
Peter Eisentraut [Sat, 29 Oct 2005 00:41:37 +0000 (00:41 +0000)]
Translation updates

19 years agoMessage corrections
Peter Eisentraut [Sat, 29 Oct 2005 00:31:52 +0000 (00:31 +0000)]
Message corrections

19 years agoUpdate supported-platforms list.
Tom Lane [Fri, 28 Oct 2005 20:01:40 +0000 (20:01 +0000)]
Update supported-platforms list.

19 years agoFix broken markup.
Tom Lane [Fri, 28 Oct 2005 20:00:03 +0000 (20:00 +0000)]
Fix broken markup.

19 years agoReorder code so that we don't have to hold a critical section while
Tom Lane [Fri, 28 Oct 2005 19:00:19 +0000 (19:00 +0000)]
Reorder code so that we don't have to hold a critical section while
reserving SLRU space for a new MultiXact.  The original coding would have
treated out-of-disk-space as a PANIC condition, which is unnecessary.

19 years agoClean up AIX build to avoid 'duplicate symbol' warnings, by moving use
Tom Lane [Fri, 28 Oct 2005 17:32:22 +0000 (17:32 +0000)]
Clean up AIX build to avoid 'duplicate symbol' warnings, by moving use
of postgres.imp file into BE_DLLLIBS macro.  This makes the AIX build
work more like the Windows and Darwin builds, which have similar requirements
to mention a backend library when linking shared libraries that will be
dynamically loaded into the backend.

19 years agoFix race condition in multixact code: it's possible to try to read a
Tom Lane [Fri, 28 Oct 2005 17:27:29 +0000 (17:27 +0000)]
Fix race condition in multixact code: it's possible to try to read a
multixact's starting offset before the offset has been stored into the
SLRU file.  A simple fix would be to hold the MultiXactGenLock until the
offset has been stored, but that looks like a big concurrency hit.  Instead
rely on knowledge that unset offsets will be zero, and loop when we see
a zero.  This requires a little extra hacking to ensure that zero is never
a valid value for the offset.  Problem reported by Matteo Beccati, fix
ideas from Martijn van Oosterhout, Alvaro Herrera, and Tom Lane.

19 years ago1 Fix problem with lost precision in rank with OR-ed lexemes
Teodor Sigaev [Fri, 28 Oct 2005 13:05:06 +0000 (13:05 +0000)]
1 Fix problem with lost precision in rank with OR-ed lexemes
2 Allow tsquery_in to input void tsquery: resolve dump/restore problem with tsquery

19 years agoRemove mention that \x doesn't affect backslash commands.
Bruce Momjian [Fri, 28 Oct 2005 02:39:07 +0000 (02:39 +0000)]
Remove mention that \x doesn't affect backslash commands.

19 years agoAdd an ifneq to avoid make warning on AIX --- there is a separate rule
Tom Lane [Thu, 27 Oct 2005 20:45:29 +0000 (20:45 +0000)]
Add an ifneq to avoid make warning on AIX --- there is a separate rule
for postgres: on line 86, and line 43 shouldn't be used.  Noted while
looking at kookaburra buildfarm results.

19 years agoTweak buffer manager so that 'internal' accesses to a buffer do not
Tom Lane [Thu, 27 Oct 2005 17:07:58 +0000 (17:07 +0000)]
Tweak buffer manager so that 'internal' accesses to a buffer do not
advance its usage_count.  This includes writes of dirty buffers triggered
by bgwriter, checkpoint, or FlushRelationBuffers, as well as various
corner cases that really ought not count as accesses to the page.
Should make for some marginal improvement in the quality of our decisions
about when to recycle buffers.  Per suggestion from ITAGAKI Takahiro.

19 years agoUpdate:
Bruce Momjian [Thu, 27 Oct 2005 14:16:05 +0000 (14:16 +0000)]
Update:

<  o Allow ALTER TABLE RENAME CONSTRAINT
>  o Add ALTER TABLE RENAME CONSTRAINT, update index name also

19 years agoAdd:
Bruce Momjian [Thu, 27 Oct 2005 14:07:29 +0000 (14:07 +0000)]
Add:

>  o Have ALTER INDEX update the name of a constraint using that index
>  o Allow ALTER TABLE RENAME CONSTRAINT

19 years agoAdd mention that .pgpass localhost matches tcp and unix domain sockets.
Bruce Momjian [Thu, 27 Oct 2005 13:53:41 +0000 (13:53 +0000)]
Add mention that .pgpass localhost matches tcp and unix domain sockets.

19 years agoAdd mention of errfinish.
Bruce Momjian [Thu, 27 Oct 2005 13:48:14 +0000 (13:48 +0000)]
Add mention of errfinish.

19 years agoDisable expanded mode only for \d tablename, not for all backslash
Bruce Momjian [Thu, 27 Oct 2005 13:34:47 +0000 (13:34 +0000)]
Disable expanded mode only for \d tablename, not for all backslash
commands.  Per complaint that \df+ is clearer in expanded mode.

19 years agoAdd comment documenting actual failure case of using
Bruce Momjian [Thu, 27 Oct 2005 02:45:22 +0000 (02:45 +0000)]
Add comment documenting actual failure case of using
interval_justify_hours in timestamp subtraction.  TODO already has text
description.

19 years agoAdd items from Tom:
Bruce Momjian [Thu, 27 Oct 2005 02:33:31 +0000 (02:33 +0000)]
Add items from Tom:

<  o Add a GUC variable to allow output of interval values in ISO8601
<    format
212a211,223
>  o Add a GUC variable to allow output of interval values in ISO8601
>    format
>  o Improve timestamptz subtraction to be DST-aware
>
>    Currently, subtracting one date from another that crosses a
>    daylight savings time adjustment can return '1 day 1 hour', but
>    adding that back to the first date returns a time one hour in
>    the future.  This is caused by the adjustment of '25 hours' to
>    '1 day 1 hour', and '1 day' is the same time the next day, even
>    if daylight savings adjustments are involved.
>
>  o Fix interval display to support values exceeding 2^31 hours
>  o Add overflow checking to timestamp and interval arithmetic

19 years agoAdd info about using rsync to make base backups. Per report from
Tom Lane [Wed, 26 Oct 2005 20:42:35 +0000 (20:42 +0000)]
Add info about using rsync to make base backups.  Per report from
Jeff Frost, it may be necessary to ignore specific exit codes.

19 years agoAdjust parser so that POSTQUEL-style implicit RTEs are stored with
Tom Lane [Wed, 26 Oct 2005 19:21:55 +0000 (19:21 +0000)]
Adjust parser so that POSTQUEL-style implicit RTEs are stored with
inFromCl true, meaning that they will list out as explicit RTEs if they
are in a view or rule.  Update comments about inFromCl to reflect the way
it's now actually used.  Per recent discussion.

19 years agoRename config section headings.
Bruce Momjian [Wed, 26 Oct 2005 12:55:07 +0000 (12:55 +0000)]
Rename config section headings.

Simon Riggs

19 years agoFix longstanding bug that would sometimes let the planner generate a bad plan
Tom Lane [Tue, 25 Oct 2005 20:30:30 +0000 (20:30 +0000)]
Fix longstanding bug that would sometimes let the planner generate a bad plan
for an outer join; symptom is bogus error "RIGHT JOIN is only supported with
merge-joinable join conditions".  Problem was that select_mergejoin_clauses
did its tests in the wrong order.  We need to force left join not right join
for a merge join when there are non-mergeable join clauses; but the test for
this only accounted for mergejoinability of the clause operator, and not
whether the left and right Vars were of the proper relations.  Per report
from Jean-Pierre Pelletier.

19 years agoAdd:
Bruce Momjian [Tue, 25 Oct 2005 18:50:56 +0000 (18:50 +0000)]
Add:

>
>  o Add auto-expanded mode so expanded output is used if the row
>    length is wider than the screen width.
>
>    Consider using auto-expanded mode for backslash commands like \df+.

19 years agoConvert explanation of interval change into something resembling grammatical
Tom Lane [Tue, 25 Oct 2005 17:54:30 +0000 (17:54 +0000)]
Convert explanation of interval change into something resembling grammatical
English, and fix the incorrect example.

19 years agoRemove justify_hours call from interval_mul and interval_div, and make
Tom Lane [Tue, 25 Oct 2005 17:13:07 +0000 (17:13 +0000)]
Remove justify_hours call from interval_mul and interval_div, and make
some small stylistic improvements in these functions.  Also fix several
places where TMODULO() was being used with wrong-sized quotient argument,
creating a risk of overflow --- interval2tm was actually capable of going
into an infinite loop because of this.

19 years agoSmall grammar correction.
Tom Lane [Tue, 25 Oct 2005 15:47:51 +0000 (15:47 +0000)]
Small grammar correction.

19 years agoFix Windows setitimer() emulation to not depend on delivering an APC
Tom Lane [Tue, 25 Oct 2005 15:15:16 +0000 (15:15 +0000)]
Fix Windows setitimer() emulation to not depend on delivering an APC
to the main thread.  This allows removal of WaitForSingleObjectEx() calls
from the main thread, thereby allowing us to re-enable Qingqing Zhou's
CHECK_FOR_INTERRUPTS performance improvement.  Qingqing, Magnus, et al.

19 years agoUpdate 24 hours/1 day distinction in release notes.
Bruce Momjian [Tue, 25 Oct 2005 15:12:22 +0000 (15:12 +0000)]
Update 24 hours/1 day distinction in release notes.

19 years agoChange trace_sort to output to the log, rather than the user's terminal.
Bruce Momjian [Tue, 25 Oct 2005 13:47:08 +0000 (13:47 +0000)]
Change trace_sort to output to the log, rather than the user's terminal.

19 years agoUpdate the URL for "The Hitch-Hiker's Guide to Evolutionary Computation".
Bruce Momjian [Tue, 25 Oct 2005 13:38:09 +0000 (13:38 +0000)]
Update the URL for "The Hitch-Hiker's Guide to Evolutionary Computation".

Sergey E. Koposov

19 years agoIBM has addressed the socket address storage issue as of AIX 5.3
Bruce Momjian [Mon, 24 Oct 2005 22:30:35 +0000 (22:30 +0000)]
IBM has addressed the socket address storage issue as of AIX 5.3
maintenance level 5300-03; the following patch adds documentation to
FAQ_AIX.

Chris Browne

19 years agoAdd:
Bruce Momjian [Mon, 24 Oct 2005 18:08:52 +0000 (18:08 +0000)]
Add:

> * Allow user-defined functions retuning a domain value to enforce domain
>   constraints

19 years agoMore forcefully recommend MD5 over crypt authentication.
Bruce Momjian [Mon, 24 Oct 2005 15:49:54 +0000 (15:49 +0000)]
More forcefully recommend MD5 over crypt authentication.

19 years agoFix incorrect wording about function failure time on unsafe ops - these
Andrew Dunstan [Mon, 24 Oct 2005 15:39:50 +0000 (15:39 +0000)]
Fix incorrect wording about function failure time on unsafe ops - these
are now caught by the validator. And a small visit from the perl style police:
check the return value from open().

19 years agoI have applied the following patch to document PQinitSSL() and
Bruce Momjian [Mon, 24 Oct 2005 15:38:37 +0000 (15:38 +0000)]
I have applied the following patch to document PQinitSSL() and
PQregisterThreadLock().

I also remove the crypt() mention in the libpq threading section and
added a single sentence in the client-auth manual page under crypt().
Crypt authentication is so old now that a separate paragraph about it
seemed unwise.

I also added a comment about our use of locking around pqGetpwuid().

19 years agoEnsure that a plpgsql LOOP with an empty body still executes at least
Tom Lane [Mon, 24 Oct 2005 15:10:22 +0000 (15:10 +0000)]
Ensure that a plpgsql LOOP with an empty body still executes at least
one CHECK_FOR_INTERRUPTS() call, so that you can control-C out of the
loop.  Reported by Merlin Moncure.

19 years agoA bit of minor copy-editing.
Tom Lane [Sun, 23 Oct 2005 19:29:49 +0000 (19:29 +0000)]
A bit of minor copy-editing.

19 years agoI've confirmed 8.1beta4 passes regression tests on all hardware platforms
Tom Lane [Sun, 23 Oct 2005 18:30:49 +0000 (18:30 +0000)]
I've confirmed 8.1beta4 passes regression tests on all hardware platforms
of RHEL 4.  Update supported-platforms list accordingly.

19 years agoupdate configure and bugtemplate for beta 4 ... REL8_1_0BETA4
PostgreSQL Daemon [Sat, 22 Oct 2005 22:46:33 +0000 (22:46 +0000)]
update configure and bugtemplate for beta 4 ...

19 years agoAdd a note about GNU tar's propensity to complain if a file changes
Tom Lane [Sat, 22 Oct 2005 22:09:49 +0000 (22:09 +0000)]
Add a note about GNU tar's propensity to complain if a file changes
while tar is copying it.  This behavior is unhelpful when taking a base
backup.  Per gripe from Pallav Kalva back in April.

19 years agoCopy-editing for recent documentation changes relevant to WAL,
Tom Lane [Sat, 22 Oct 2005 21:56:07 +0000 (21:56 +0000)]
Copy-editing for recent documentation changes relevant to WAL,
full_page_writes, etc.

19 years agoMake code for selecting default WAL sync method less confusing.
Tom Lane [Sat, 22 Oct 2005 20:27:17 +0000 (20:27 +0000)]
Make code for selecting default WAL sync method less confusing.

19 years agoFix documentation to specify the correct range of timezone offsets for
Tom Lane [Sat, 22 Oct 2005 19:33:57 +0000 (19:33 +0000)]
Fix documentation to specify the correct range of timezone offsets for
type time with time zone, ie, +/- 13:59 not +/- 12:00.  Also some minor
wording improvements.

19 years agoTemporarily disable Qingqing's Windows signal processing patch, so that
Tom Lane [Sat, 22 Oct 2005 17:09:48 +0000 (17:09 +0000)]
Temporarily disable Qingqing's Windows signal processing patch, so that
WaitForSingleObjectEx is always called by CHECK_FOR_INTERRUPTS.  This
should be reinstated but the setitimer() emulation will have to be
redesigned first.

19 years agoFix typo.
Alvaro Herrera [Sat, 22 Oct 2005 14:44:35 +0000 (14:44 +0000)]
Fix typo.

19 years agominor code cleanup - replace useless struct timezone argument to
Andrew Dunstan [Sat, 22 Oct 2005 14:27:29 +0000 (14:27 +0000)]
minor code cleanup - replace useless struct timezone argument to
gettimeofday with NULL in a few places, making it consistent with
usage elsewhere.

19 years agoImprove performance of CHECK_FOR_INTERRUPTS() macro on Windows by not doing
Tom Lane [Fri, 21 Oct 2005 21:43:46 +0000 (21:43 +0000)]
Improve performance of CHECK_FOR_INTERRUPTS() macro on Windows by not doing
a kernel call unless there's some evidence of a pending signal.  This should
bring its performance on Windows into line with the Unix version.  Problem
diagnosis and patch by Qingqing Zhou.  Minor stylistic tweaks by moi ...
if it's broken, it's my fault.

19 years agoClean up autovacuum documentation, which was a bit out of sync with what
Tom Lane [Fri, 21 Oct 2005 19:39:08 +0000 (19:39 +0000)]
Clean up autovacuum documentation, which was a bit out of sync with what
the code actually does, and needed copy-editing anyway.  Also take the
opportunity to expand the section on routine reindexing.

19 years agoFix EXPLAIN ANALYZE bug noted by Wiebe Cazemier: although we were
Tom Lane [Fri, 21 Oct 2005 16:43:33 +0000 (16:43 +0000)]
Fix EXPLAIN ANALYZE bug noted by Wiebe Cazemier: although we were
properly advancing the CommandCounter between multiple sub-queries
generated by rules, we forgot to update the snapshot being used, so
that the successive sub-queries didn't actually see each others'
results.  This is still not *exactly* like the semantics of normal
execution of the same queries, in that we don't take new transaction
snapshots and hence don't see changes from concurrently committed
commands, but I think that's OK and probably even preferable for
EXPLAIN ANALYZE.

19 years agoAdd an implicit cast from varchar to regclass, so that existing code
Tom Lane [Fri, 21 Oct 2005 15:45:06 +0000 (15:45 +0000)]
Add an implicit cast from varchar to regclass, so that existing code
of the form nextval('foo'::varchar) doesn't break.  Per gripe from
Jean-Pierre Pelletier.  Initdb forced :-(

19 years agoAdd missing PQinitSSL and PQregisterThreadLock to exports.txt.
Tom Lane [Fri, 21 Oct 2005 15:21:21 +0000 (15:21 +0000)]
Add missing PQinitSSL and PQregisterThreadLock to exports.txt.

19 years agoUpdate obsolete URL, per Chris.
Tom Lane [Fri, 21 Oct 2005 13:59:05 +0000 (13:59 +0000)]
Update obsolete URL, per Chris.

19 years agoMinor PL/PgSQL doc tweak: use current_timestamp rather than now() in
Neil Conway [Fri, 21 Oct 2005 05:11:23 +0000 (05:11 +0000)]
Minor PL/PgSQL doc tweak: use current_timestamp rather than now() in
an example function.

19 years agoClean up some obsolete statements about GiST indexes, and add a section
Tom Lane [Fri, 21 Oct 2005 01:41:28 +0000 (01:41 +0000)]
Clean up some obsolete statements about GiST indexes, and add a section
documenting GiST crash recovery procedures, as requested some time ago
by Teodor.  (The GiST chapter doesn't seem quite the right place for
the latter, but I'm not sure what else to do with it.)

19 years agoImprove testlibpq3.c's example of PQexecParams() usage to include sending
Tom Lane [Thu, 20 Oct 2005 23:57:52 +0000 (23:57 +0000)]
Improve testlibpq3.c's example of PQexecParams() usage to include sending
a parameter in binary format.  Also, add a TIP explaining how to use casts
in the query text to avoid needing to specify parameter types by OID.
Also fix bogus spacing --- apparently somebody expanded the tabs in the
example programs to 8 spaces instead of 4 when transposing them into SGML.

19 years agoMinor tweak to libpq documentation: make "PREPARE" and "DEALLOCATE" xrefs.
Neil Conway [Thu, 20 Oct 2005 21:04:14 +0000 (21:04 +0000)]
Minor tweak to libpq documentation: make "PREPARE" and "DEALLOCATE" xrefs.

19 years agoPostpone pg_timezone_initialize() until after creation of postmaster.pid,
Tom Lane [Thu, 20 Oct 2005 20:05:45 +0000 (20:05 +0000)]
Postpone pg_timezone_initialize() until after creation of postmaster.pid,
since it can take a fair amount of time and this can confuse boot scripts
that expect postmaster.pid to appear quickly.  Move initialization of SSL
library and preloaded libraries to after that point, too, just for luck.
Per reports from Tony Caduto and others.

19 years agoDocument the behavior of GRANT/REVOKE in cases where the privilege is
Tom Lane [Thu, 20 Oct 2005 19:18:01 +0000 (19:18 +0000)]
Document the behavior of GRANT/REVOKE in cases where the privilege is
held by means of role membership, rather than directly.  Per discussion
and bug fix of a couple weeks ago.

19 years agoAdjust not-too-sane calculation of DDD value for to_char(interval).
Tom Lane [Thu, 20 Oct 2005 15:59:46 +0000 (15:59 +0000)]
Adjust not-too-sane calculation of DDD value for to_char(interval).
Per gripe from Chris Matheson.

19 years agoClean up md5.c to make it clearer that it is a frontend-and-backend
Tom Lane [Thu, 20 Oct 2005 13:54:08 +0000 (13:54 +0000)]
Clean up md5.c to make it clearer that it is a frontend-and-backend
module.  Don't rely on backend palloc semantics; in fact, best to not
use palloc at all, rather than #define'ing it to malloc, because that
just encourages errors of omission.  Bug spotted by Volkan YAZICI,
but I went further than he did to fix it.

19 years agoMake \d order a table's check constraints by constraint name instead
Tom Lane [Thu, 20 Oct 2005 05:15:09 +0000 (05:15 +0000)]
Make \d order a table's check constraints by constraint name instead
of the text of the constraint condition.  Per Chris K-L, though I didn't
use his patch exactly.

19 years agoBetter solution to the problem of labeling whole-row Datums that are
Tom Lane [Wed, 19 Oct 2005 22:30:30 +0000 (22:30 +0000)]
Better solution to the problem of labeling whole-row Datums that are
generated from subquery outputs: use the type info stored in the Var
itself.  To avoid making ExecEvalVar and slot_getattr more complex
and slower, I split out the whole-row case into a separate ExecEval routine.

19 years agoEnsure that the Datum generated from a whole-row Var contains valid
Tom Lane [Wed, 19 Oct 2005 18:18:33 +0000 (18:18 +0000)]
Ensure that the Datum generated from a whole-row Var contains valid
type ID information even when it's a record type.  This is needed to
handle whole-row Vars referencing subquery outputs.  Per example from
Richard Huxton.

19 years agoFix oversight in recent changes to enable the 'physical tlist'
Tom Lane [Wed, 19 Oct 2005 17:31:20 +0000 (17:31 +0000)]
Fix oversight in recent changes to enable the 'physical tlist'
optimization for subquery and function scan nodes: we can't just do it
unconditionally, we still have to check whether there is any need for
a whole-row Var.  I had been thinking that these node types couldn't
have any system columns, which is true, but that loop is also checking
for attno zero, ie, whole-row Var.  Fix comment to not be so misleading.
Per test case from Richard Huxton.

19 years agoImprove trace_sort code to also show the total memory or disk space used.
Tom Lane [Tue, 18 Oct 2005 22:59:37 +0000 (22:59 +0000)]
Improve trace_sort code to also show the total memory or disk space used.
Per request from Marc.

19 years agoreplace use of predefined perl vars $a and $b with $x and $y - per Greg Sabino Mullane
Andrew Dunstan [Tue, 18 Oct 2005 22:53:54 +0000 (22:53 +0000)]
replace use of predefined perl vars $a and $b with $x and $y - per Greg Sabino Mullane

19 years agoAdd an entry to the discussion of regression test failures about the
Tom Lane [Tue, 18 Oct 2005 21:43:33 +0000 (21:43 +0000)]
Add an entry to the discussion of regression test failures about the
possibility of a failure due to stack overflow when max_stack_depth is
not set properly for the platform.

19 years agoCode review for regexp_replace patch. Improve documentation and comments,
Tom Lane [Tue, 18 Oct 2005 20:38:58 +0000 (20:38 +0000)]
Code review for regexp_replace patch.  Improve documentation and comments,
fix problems with replacement-string backslashes that aren't followed by
one of the expected characters, avoid giving the impression that
replace_text_regexp() is meant to be called directly as a SQL function,
etc.

19 years agoCode review for spi_query/spi_fetchrow patch: handle errors sanely,
Tom Lane [Tue, 18 Oct 2005 17:13:14 +0000 (17:13 +0000)]
Code review for spi_query/spi_fetchrow patch: handle errors sanely,
avoid leaking memory.  I would add a regression test for error handling
except it seems eval{} can't be used in unprivileged plperl :-(