]> granicus.if.org Git - postgresql/log
postgresql
23 years agoCOMMIT_DELAY and COMMIT_SIBLINGS were missing from the list of WAL
Tom Lane [Fri, 4 May 2001 23:11:37 +0000 (23:11 +0000)]
COMMIT_DELAY and COMMIT_SIBLINGS were missing from the list of WAL
configuration parameters in chapter 3.

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 4 May 2001 22:08:53 +0000 (22:08 +0000)]
Update TODO list.

23 years agoMakefile should have automatic dependency for parser.o too, if it's
Tom Lane [Fri, 4 May 2001 22:01:03 +0000 (22:01 +0000)]
Makefile should have automatic dependency for parser.o too, if it's
going to have any at all.

23 years agoUpdate TODO list.
Bruce Momjian [Fri, 4 May 2001 20:48:14 +0000 (20:48 +0000)]
Update TODO list.

23 years agoSeems like we should not hold off cancel/die interrupts while we are
Tom Lane [Fri, 4 May 2001 18:39:16 +0000 (18:39 +0000)]
Seems like we should not hold off cancel/die interrupts while we are
running deferred triggers.  They are really part of the regular
transaction, and they could take awhile.

23 years agoAdd missing tag.
Bruce Momjian [Fri, 4 May 2001 02:54:33 +0000 (02:54 +0000)]
Add missing tag.

23 years agoConsolidate several near-identical uses of mktime() into a single
Tom Lane [Thu, 3 May 2001 22:53:07 +0000 (22:53 +0000)]
Consolidate several near-identical uses of mktime() into a single
routine DetermineLocalTimeZone().  In that routine, be more wary of
broken mktime() implementations than the original code was: don't allow
mktime to change the already-set y/m/d/h/m/s information, and don't
use tm_gmtoff if mktime failed.  Possibly this will resolve some of
the complaints we've been hearing from users of Middle Eastern timezones
on RedHat.

23 years agoBTW it does not add encodign it just patches existing one (KOI8) to
Bruce Momjian [Thu, 3 May 2001 21:38:45 +0000 (21:38 +0000)]
BTW it does not add encodign it just patches existing one (KOI8) to
support two - KOI8-R and KOI8-U (latter is superset of the former if
not to take to the account pseudographics)

Andy Rysin

23 years agoPermission checking wasn't quite right for insert/update/delete rules,
Tom Lane [Thu, 3 May 2001 21:16:48 +0000 (21:16 +0000)]
Permission checking wasn't quite right for insert/update/delete rules,
either :-(.

23 years agopg_dump can dump 7.0 databases.
Bruce Momjian [Thu, 3 May 2001 20:39:52 +0000 (20:39 +0000)]
pg_dump can dump 7.0 databases.

23 years agoapply a little patch:
Bruce Momjian [Thu, 3 May 2001 19:20:24 +0000 (19:20 +0000)]
apply a little patch:

Oleg Bartunov

23 years agoEnsure that btree sort ordering functions and boolean comparison operators
Tom Lane [Thu, 3 May 2001 19:00:37 +0000 (19:00 +0000)]
Ensure that btree sort ordering functions and boolean comparison operators
give consistent results for all datatypes.  Types float4, float8, and
numeric were broken for NaN values; abstime, timestamp, and interval
were broken for INVALID values; timetz was just plain broken (some
possible pairs of values were neither < nor = nor >).  Also clean up
text, bpchar, varchar, and bit/varbit to eliminate duplicate code and
thereby reduce the probability of similar inconsistencies arising in
the future.

23 years agoDescribe handling of multiply-inherited fields correctly.
Tom Lane [Thu, 3 May 2001 17:50:55 +0000 (17:50 +0000)]
Describe handling of multiply-inherited fields correctly.

23 years agoPermissions were not checked correctly when one view invokes another.
Tom Lane [Thu, 3 May 2001 17:47:49 +0000 (17:47 +0000)]
Permissions were not checked correctly when one view invokes another.
Per bug report from Lieven Van Acker, 5/2/01.

23 years agoUpdate shared memory docs.
Bruce Momjian [Thu, 3 May 2001 17:28:13 +0000 (17:28 +0000)]
Update shared memory docs.

23 years agoImprove list.
Bruce Momjian [Thu, 3 May 2001 17:20:14 +0000 (17:20 +0000)]
Improve list.

23 years agoFinal stamp for 7.1.1. All done.
Bruce Momjian [Thu, 3 May 2001 17:17:41 +0000 (17:17 +0000)]
Final stamp for 7.1.1.  All done.

23 years agoUpdates for 7.1.1. Not done yet.
Bruce Momjian [Thu, 3 May 2001 16:47:58 +0000 (16:47 +0000)]
Updates for 7.1.1.  Not done yet.

23 years agoautoconf
Bruce Momjian [Thu, 3 May 2001 16:43:26 +0000 (16:43 +0000)]
autoconf

23 years agoSome updates for 7.1.1.
Bruce Momjian [Thu, 3 May 2001 16:26:21 +0000 (16:26 +0000)]
Some updates for 7.1.1.

23 years agoCorrect pg_description entry for type macaddr.
Tom Lane [Thu, 3 May 2001 16:18:02 +0000 (16:18 +0000)]
Correct pg_description entry for type macaddr.

23 years agoFixes to make ecpg work on Cygwin, from Jason Tishler <Jason.Tishler@dothill.com>.
Tom Lane [Thu, 3 May 2001 16:07:52 +0000 (16:07 +0000)]
Fixes to make ecpg work on Cygwin, from Jason Tishler <Jason.Tishler@dothill.com>.

23 years agoInitial checkin of a contributed type that handles passwords efficiently.
D'Arcy J.M. Cain [Thu, 3 May 2001 12:32:13 +0000 (12:32 +0000)]
Initial checkin of a contributed type that handles passwords efficiently.

23 years agoIn the outputs section the NOTICE reads:
Bruce Momjian [Wed, 2 May 2001 20:35:21 +0000 (20:35 +0000)]
In the outputs section the NOTICE reads:
aggregate 'agg' for 'type' does not exist.

Mismatch with synopsis, agg must be name here.

Frank Wegmann

23 years agoAdded mention of SHMALL and removed BSD/OS-specific table.
Bruce Momjian [Wed, 2 May 2001 17:04:53 +0000 (17:04 +0000)]
Added mention of SHMALL and removed BSD/OS-specific table.

23 years agodescription of the command line option echo should read:
Bruce Momjian [Wed, 2 May 2001 14:46:33 +0000 (14:46 +0000)]
description of the command line option echo should read:

Echo the queries that createuser generates and sends to the backend.
                      ^^^^^^^^^^

Frank Wegmann

23 years agoPrevious commit mistakenly converted all newlines to DOS style (CR/LF).
Tom Lane [Wed, 2 May 2001 14:23:40 +0000 (14:23 +0000)]
Previous commit mistakenly converted all newlines to DOS style (CR/LF).
Convert back to Unix style --- it seems some versions of nmake insist
on this.

23 years agoAdd note explaining why inserts take longer as tables grow. Also suggest
D'Arcy J.M. Cain [Wed, 2 May 2001 11:21:57 +0000 (11:21 +0000)]
Add note explaining why inserts take longer as tables grow.  Also suggest
the way to handle this.

23 years agoChange "|zzlzzzz" argument specification to "|zzizzzz" so that the code works
D'Arcy J.M. Cain [Wed, 2 May 2001 11:17:24 +0000 (11:17 +0000)]
Change "|zzlzzzz" argument specification to "|zzizzzz" so that the code works
properly on 64 bit systems.

Change submitted by Marc Poinot (Marc.Poinot@onera.fr)

23 years agoBump version to 7.1.1. Now run over to hub to rebuild documentation
Peter Eisentraut [Tue, 1 May 2001 10:50:26 +0000 (10:50 +0000)]
Bump version to 7.1.1.  Now run over to hub to rebuild documentation
before release...

23 years agoMore markup fixes
Peter Eisentraut [Tue, 1 May 2001 10:48:57 +0000 (10:48 +0000)]
More markup fixes

23 years agoFix slightly confused description of debug options.
Tom Lane [Tue, 1 May 2001 03:00:28 +0000 (03:00 +0000)]
Fix slightly confused description of debug options.

23 years agoSynced gram.y and preproc.y.
Tom Lane [Tue, 1 May 2001 02:33:55 +0000 (02:33 +0000)]
Synced gram.y and preproc.y.

23 years agoAllow a string argument to the EXTRACT() function.
Thomas G. Lockhart [Tue, 1 May 2001 01:36:10 +0000 (01:36 +0000)]
Allow a string argument to the EXTRACT() function.
 This is an extension to the SQL9x standard, but is consistant with usage
 of the underlying date_part() function used to implement it.
 Example: EXTRACT('YEAR',...)
No impact on regression tests.

23 years ago1) fix bugs reported by Andrea Aime.
Hiroshi Inoue [Tue, 1 May 2001 00:47:06 +0000 (00:47 +0000)]
1) fix bugs reported by Andrea Aime.
2) fix a bug reported by Jan Wieck.
psqlodbc is 7.01.0005 now.

23 years agoUpdate FAQ_HPUX now that we have reports of out-of-the-box build
Tom Lane [Mon, 30 Apr 2001 23:55:26 +0000 (23:55 +0000)]
Update FAQ_HPUX now that we have reports of out-of-the-box build
on HPUX 11.

23 years agoDocument pltcl's return_null command.
Tom Lane [Mon, 30 Apr 2001 22:22:34 +0000 (22:22 +0000)]
Document pltcl's return_null command.

23 years agoRepair freshly-broken markup.
Tom Lane [Mon, 30 Apr 2001 22:21:22 +0000 (22:21 +0000)]
Repair freshly-broken markup.

23 years agoexec_move_row() should be more forgiving of tuples with a different
Tom Lane [Mon, 30 Apr 2001 20:05:40 +0000 (20:05 +0000)]
exec_move_row() should be more forgiving of tuples with a different
number of columns than it was expecting, for reasons that are now
documented in the code...

23 years agoSuppress pull-up of subqueries that are in the nullable side of an outer
Tom Lane [Mon, 30 Apr 2001 19:24:47 +0000 (19:24 +0000)]
Suppress pull-up of subqueries that are in the nullable side of an outer
join.  This is needed to avoid improper evaluation of expressions that
should be nulled out, as in Victor Wagner's bug report of 4/27/01.
Pretty ugly solution, but no time to do anything better for 7.1.1.

23 years agoBreak up PQexec() result functions into subsections to be clearer. Both
Bruce Momjian [Mon, 30 Apr 2001 17:38:00 +0000 (17:38 +0000)]
Break up PQexec() result functions into subsections to be clearer.  Both
libpq and libpq++ reorganized.

23 years agoUpdate TODO list.
Bruce Momjian [Mon, 30 Apr 2001 15:57:01 +0000 (15:57 +0000)]
Update TODO list.

23 years agoAdd mention of libpgtcl's options to manual, libpq++ markup fixes.
Bruce Momjian [Mon, 30 Apr 2001 04:26:01 +0000 (04:26 +0000)]
Add mention of libpgtcl's options to manual, libpq++ markup fixes.

23 years agoUpdate TODO list.
Bruce Momjian [Mon, 30 Apr 2001 02:35:40 +0000 (02:35 +0000)]
Update TODO list.

23 years agoAdd missing Unicode support for Cyrillic encodings.
Tatsuo Ishii [Sun, 29 Apr 2001 07:27:38 +0000 (07:27 +0000)]
Add missing Unicode support for Cyrillic encodings.
Patches contributed by Victor Wagner.

23 years agoIn create_function there is a remark concerning SQL/PSM not
Bruce Momjian [Sat, 28 Apr 2001 13:59:07 +0000 (13:59 +0000)]
In create_function there is a remark concerning SQL/PSM not
being ratified as yet. This is certainly no longer true, it wasn't
even true in Q2/1998 when I did a little research for Date's book.
SQL/PSM had been published on 1996-12-15 as ISO/IEC 9075:4. So you
might want to update that section.

Frank Wegmann

23 years agoA small patch for a small option (for rel. 7.1).
Bruce Momjian [Fri, 27 Apr 2001 01:31:56 +0000 (01:31 +0000)]
A small patch for a small option (for rel. 7.1).

Jimmy Olgeni

23 years agoUpdate FAQ.
Bruce Momjian [Fri, 27 Apr 2001 00:55:49 +0000 (00:55 +0000)]
Update FAQ.

23 years agoUpdate FAQ.
Bruce Momjian [Fri, 27 Apr 2001 00:22:59 +0000 (00:22 +0000)]
Update FAQ.

23 years agoUpdate FAQ.
Bruce Momjian [Fri, 27 Apr 2001 00:21:44 +0000 (00:21 +0000)]
Update FAQ.

23 years agoRemove bogus Assert from AbsoluteTimeIsBefore(). (If you don't think
Tom Lane [Thu, 26 Apr 2001 21:52:17 +0000 (21:52 +0000)]
Remove bogus Assert from AbsoluteTimeIsBefore().  (If you don't think
it's bogus, try building a btree index on the regress tests' abstime_tbl.)

23 years agoTweak nestloop costing to weight restart cost of inner path more heavily.
Tom Lane [Wed, 25 Apr 2001 22:04:37 +0000 (22:04 +0000)]
Tweak nestloop costing to weight restart cost of inner path more heavily.
Without this, it was making some pretty silly decisions about whether an
expensive sub-SELECT should be the inner or outer side of a join...

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 25 Apr 2001 17:49:29 +0000 (17:49 +0000)]
Update TODO list.

23 years ago- Fixed CONSTRAINT TRIGGER dump to record tgconstrelid properly
Philip Warner [Wed, 25 Apr 2001 07:03:20 +0000 (07:03 +0000)]
- Fixed CONSTRAINT TRIGGER dump to record tgconstrelid properly
- pgsql v7.0 compatbility

23 years agotypological errors
Peter Eisentraut [Tue, 24 Apr 2001 20:05:02 +0000 (20:05 +0000)]
typological errors

23 years agoUndo pgindent brain damage, so that node type numbers can once again be
Tom Lane [Tue, 24 Apr 2001 00:08:38 +0000 (00:08 +0000)]
Undo pgindent brain damage, so that node type numbers can once again be
counted off by lines ...

23 years agoBehave sanely when database name is omitted from command line.
Tom Lane [Mon, 23 Apr 2001 23:36:33 +0000 (23:36 +0000)]
Behave sanely when database name is omitted from command line.

23 years agoSend 'cd' output to /dev/null.
Peter Eisentraut [Mon, 23 Apr 2001 20:27:55 +0000 (20:27 +0000)]
Send 'cd' output to /dev/null.

23 years agoDefend against ancient versions of zlib by checking that zlib.h
Tom Lane [Mon, 23 Apr 2001 15:14:58 +0000 (15:14 +0000)]
Defend against ancient versions of zlib by checking that zlib.h
defines z_streamp.

23 years agocompatible_oper needs to do ReleaseSysCache in one path to avoid
Tom Lane [Mon, 23 Apr 2001 04:32:30 +0000 (04:32 +0000)]
compatible_oper needs to do ReleaseSysCache in one path to avoid
complaints about 'Cache reference leak'.  Per report from Don Baccus.

23 years ago1) Decrease the size of needlessly large buffers. For example, it
Hiroshi Inoue [Mon, 23 Apr 2001 01:41:06 +0000 (01:41 +0000)]
1) Decrease the size of needlessly large buffers. For example, it
   resolved the stack over flow errors reported by Johann Zuschlag.
2) Support {oj syntax for 71. servers.

23 years agoA patch to fix the following bugs.
Hiroshi Inoue [Mon, 23 Apr 2001 01:00:49 +0000 (01:00 +0000)]
A patch to fix the following bugs.
1) [ODBC] Psqlodbc and Centura: here it is a patch
        posted by Matteo Cavalleli
2) [ODBC] pgsqODBC binding parameters II
        posted by Ludek Finstrle
3) Invalid Page Fault in PSQLODBC.DLL
        personal mail from Johann Zuschlag

   Hiroki Kataoka   kataoka@interwiz.koganei.tokyo.jp

23 years agoDon't bomb out on indexes on system attributes other than 'oid'.
Tom Lane [Sun, 22 Apr 2001 21:34:13 +0000 (21:34 +0000)]
Don't bomb out on indexes on system attributes other than 'oid'.
(Not sure such an index is actually useful, but just because it's
useless doesn't mean pg_dump should coredump.)

23 years agoRemove useless setuid() call, instead add a check that real and effective
Tom Lane [Sat, 21 Apr 2001 18:29:29 +0000 (18:29 +0000)]
Remove useless setuid() call, instead add a check that real and effective
userids are the same.  Per today's pghackers discussion.

23 years agoMake sure that the postmaster options (from postmaster.opts[.default] or
Peter Eisentraut [Sat, 21 Apr 2001 11:23:58 +0000 (11:23 +0000)]
Make sure that the postmaster options (from postmaster.opts[.default] or
-o option) are properly dequoted.  Also, always pass an explicit -D option
to postmaster, don't rely on it being set in postmaster.opts.

23 years agoFinal markup for 7.1 hardcopy release. Mostly typo and table-fitting
Thomas G. Lockhart [Fri, 20 Apr 2001 15:52:33 +0000 (15:52 +0000)]
Final markup for 7.1 hardcopy release. Mostly typo and table-fitting
 adjustments. Note that many tables are being abused with *really* long
 description columns. Should probably shrink those columns to be more
 concise, and move some of the info to follow-on reference notes.

23 years agoAdd option --refentry to change \keepn (keep with next paragraph) to
Thomas G. Lockhart [Fri, 20 Apr 2001 15:50:25 +0000 (15:50 +0000)]
Add option --refentry to change \keepn (keep with next paragraph) to
 \keep (keep current paragraph together). This fixes most troubles with
 reference pages marked up with <refentry> tags.
Use on reference.rtf, generated by "make reference.rtf".

23 years agoRemove libsfio check.
Peter Eisentraut [Thu, 19 Apr 2001 20:27:47 +0000 (20:27 +0000)]
Remove libsfio check.

23 years agoPass -D option to pg_ctl because export PGDATA won't be preserved across
Peter Eisentraut [Thu, 19 Apr 2001 19:17:44 +0000 (19:17 +0000)]
Pass -D option to pg_ctl because export PGDATA won't be preserved across
'su -l'.

Reported by <VASQUEZ_JASON@LILLY.COM>.

23 years agoAdd extern int optreset declaration, per bug report from <info@decc.nl>.
Peter Eisentraut [Thu, 19 Apr 2001 19:09:23 +0000 (19:09 +0000)]
Add extern int optreset declaration, per bug report from <info@decc.nl>.

23 years agoFix "off by VARHDRSZ" bug in (Cyrillic) recode support.
Peter Eisentraut [Thu, 19 Apr 2001 19:01:23 +0000 (19:01 +0000)]
Fix "off by VARHDRSZ" bug in (Cyrillic) recode support.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 19 Apr 2001 15:22:29 +0000 (15:22 +0000)]
Update TODO list.

23 years agoKluge solution for Alex Pilosov's report of problems with whole-tuple
Tom Lane [Thu, 19 Apr 2001 04:29:02 +0000 (04:29 +0000)]
Kluge solution for Alex Pilosov's report of problems with whole-tuple
function arguments in join queries: copy the tuples into
TransactionCommandContext so they don't get recycled too soon.  This is
horrid, but not any worse than 7.0 or before, which also leaked such
tuples until end of query.  A proper fix will require allowing tuple
datums to be physically stored inside larger tuple datums, which opens
up a bunch of issues that can't realistically be solved for 7.1.1.

23 years agoAdd a crash gurard to pg_encoding_mblen in case of an invalid encoding
Tatsuo Ishii [Thu, 19 Apr 2001 02:34:35 +0000 (02:34 +0000)]
Add a crash gurard to pg_encoding_mblen in case of an invalid encoding
given.

23 years agoFurther tweaking of error messages for cases involving attributes &
Tom Lane [Wed, 18 Apr 2001 22:25:31 +0000 (22:25 +0000)]
Further tweaking of error messages for cases involving attributes &
functions of join or subselect aliases.  It'd be awfully nice if this
code knew for sure whether it was dealing with 'x.f' or 'f(x)' syntax;
maybe we can fix that in a future cycle.

23 years agoRepair misbehavior of \! with immediately following argument that contains
Peter Eisentraut [Wed, 18 Apr 2001 20:53:08 +0000 (20:53 +0000)]
Repair misbehavior of \! with immediately following argument that contains
spaces (e.g., '\!ls -l').  Also correct a comment.

23 years agoMake the world safe for passing whole rows of views to functions. This
Tom Lane [Wed, 18 Apr 2001 20:42:56 +0000 (20:42 +0000)]
Make the world safe for passing whole rows of views to functions.  This
already worked fine for whole rows of tables, but not so well for views...

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 18 Apr 2001 17:52:12 +0000 (17:52 +0000)]
Update TODO list.

23 years agoMake rule lister do the right thing with Vars representing whole tuples.
Tom Lane [Wed, 18 Apr 2001 17:04:24 +0000 (17:04 +0000)]
Make rule lister do the right thing with Vars representing whole tuples.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 17 Apr 2001 14:17:47 +0000 (14:17 +0000)]
Update TODO list.

23 years agoThe renewal of README thanks to Dave Page.
Hiroshi Inoue [Tue, 17 Apr 2001 07:30:53 +0000 (07:30 +0000)]
The renewal of README thanks to Dave Page.

23 years agoAdd some defenses to guard against case where a rule refers to a table
Tom Lane [Tue, 17 Apr 2001 00:32:58 +0000 (00:32 +0000)]
Add some defenses to guard against case where a rule refers to a table
or view that's been dropped and then recreated with the same name (but,
perhaps, different columns).  Eventually we'd like to support this but
for now all we can do is fail cleanly, rather than possibly coredumping
if we proceed using the obsolete rule.

23 years agoAvoid reversing user-given order of WHERE clauses while attaching clauses
Tom Lane [Mon, 16 Apr 2001 19:44:10 +0000 (19:44 +0000)]
Avoid reversing user-given order of WHERE clauses while attaching clauses
to specific base or join RelOptInfo nodes during planning.  This preserves
the more-intuitive behavior of 7.0.* --- if you write an expensive clause
(such as a sub-select) last, it should get evaluated last.  Someday we
ought to try to have some intelligence about the order of evaluation of
WHERE clauses, but for now we should not override what the user wrote.

23 years agoCorrection for mathematical properties in Unicode converison maps.
Tatsuo Ishii [Mon, 16 Apr 2001 06:10:19 +0000 (06:10 +0000)]
Correction for mathematical properties in Unicode converison maps.
Patches contributed by Eiji Tokuya (e-tokuya@sankyo-unyu.co.jp)

23 years agoUpdate Japanese FAQ's.
Bruce Momjian [Mon, 16 Apr 2001 04:20:36 +0000 (04:20 +0000)]
Update Japanese FAQ's.

23 years agoAdd new item.
Bruce Momjian [Mon, 16 Apr 2001 04:11:54 +0000 (04:11 +0000)]
Add new item.

23 years agogetdatabaseencoding() and PG_encoding_to_char() were being sloppy about
Tom Lane [Mon, 16 Apr 2001 02:42:01 +0000 (02:42 +0000)]
getdatabaseencoding() and PG_encoding_to_char() were being sloppy about
converting char* strings to type 'name'.  Imagine my surprise when 7.1
release coredumped upon start when compiled --enable-multibyte ...

23 years agoRemove error report from pq_endmessage when pq_putmessage fails. The
Tom Lane [Mon, 16 Apr 2001 01:46:57 +0000 (01:46 +0000)]
Remove error report from pq_endmessage when pq_putmessage fails.  The
only possible failure is in pq_flush, which will log a (better!) report
anyway --- so pq_endmessage is just cluttering the log with a redundant
entry.  This matters when a client crashes partway through a large query,
since we will emit many broken-pipe reports before finishing the query
and exiting.

23 years agoUpdate TODO list.
Bruce Momjian [Mon, 16 Apr 2001 01:36:50 +0000 (01:36 +0000)]
Update TODO list.

23 years agoAdd LDFLAGS_SL to AIX shlib building command, per report from
Tom Lane [Sun, 15 Apr 2001 03:25:07 +0000 (03:25 +0000)]
Add LDFLAGS_SL to AIX shlib building command, per report from
Rocco Altier <roccoa@routescape.com>.

23 years agoConstant expressions that appear in ORDER BY, GROUP BY, DISTINCT ON
Tom Lane [Sun, 15 Apr 2001 03:14:18 +0000 (03:14 +0000)]
Constant expressions that appear in ORDER BY, GROUP BY, DISTINCT ON
lists should be reverse-compiled into targetlist index numbers, because
that's the only interpretation the parser allows for a constant in these
clauses.  (Ergo, the only way they could have gotten into the list in
the first place is to have come from the targetlist; so this should always
work.)  Per problem report from Peter E.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 15 Apr 2001 01:56:08 +0000 (01:56 +0000)]
Update TODO list.

23 years agoPrevent generation of invalid plans for RIGHT or FULL joins with multiple
Tom Lane [Sun, 15 Apr 2001 00:48:17 +0000 (00:48 +0000)]
Prevent generation of invalid plans for RIGHT or FULL joins with multiple
join clauses.  The mergejoin executor wants all the join clauses to appear
as merge quals, not as extra joinquals, for these kinds of joins.  But the
planner would consider plans in which partially-sorted input paths were
used, leading to only some of the join clauses becoming merge quals.
This is fine for inner/left joins, not fine for right/full joins.

23 years agoIf the password prompt goes to stderr, then the trailing newline should
Peter Eisentraut [Sun, 15 Apr 2001 00:43:37 +0000 (00:43 +0000)]
If the password prompt goes to stderr, then the trailing newline should
also go there.

23 years agoWork with Readline 4.2.
Peter Eisentraut [Sat, 14 Apr 2001 22:55:02 +0000 (22:55 +0000)]
Work with Readline 4.2.

23 years agoAdd debug_query_string global variable for pgmonitor and debugging use.
Bruce Momjian [Sat, 14 Apr 2001 19:11:45 +0000 (19:11 +0000)]
Add debug_query_string global variable for pgmonitor and debugging use.

23 years ago- Get view OID based on rule OID not base table OID
Philip Warner [Sat, 14 Apr 2001 13:11:03 +0000 (13:11 +0000)]
- Get view OID based on rule OID not base table OID
- Fix crash due to null string pointer in some tar files with some libs

23 years agoAnother try at making numeric MODULO operator produce the right answer.
Tom Lane [Sat, 14 Apr 2001 02:10:57 +0000 (02:10 +0000)]
Another try at making numeric MODULO operator produce the right answer.
Although it was now using the right equation, it was making bogus choices
of the precision to compute intermediate results to.  I'm not sure this
is really right even yet, but it's better than before ...

23 years agoSuppress compiler warnings in Vax and NS32K assembly code: 'register foo'
Tom Lane [Fri, 13 Apr 2001 23:32:57 +0000 (23:32 +0000)]
Suppress compiler warnings in Vax and NS32K assembly code: 'register foo'
is not a complete declaration.

23 years agotag the source tree as REL7_1, renaming the old tag REL7_1_BETA like it REL7_1
Marc G. Fournier [Fri, 13 Apr 2001 21:21:33 +0000 (21:21 +0000)]
tag the source tree as REL7_1, renaming the old tag REL7_1_BETA like it
should have been done

generate a new Changelog from rc4, which is nice and short ... one python
interface change

tag configure as 7.1, for generating the packages ...

This is it folks ... Release 7.1 is officially here ...