OSDN Git Service

pg-rex/syncrep.git
20 years agoRepair RI trigger visibility problems (this time for sure ;-)) per recent
Tom Lane [Wed, 1 Oct 2003 21:30:53 +0000 (21:30 +0000)]
Repair RI trigger visibility problems (this time for sure ;-)) per recent
discussion on pgsql-hackers: in READ COMMITTED mode we just have to force
a QuerySnapshot update in the trigger, but in SERIALIZABLE mode we have
to run the scan under a current snapshot and then complain if any rows
would be updated/deleted that are not visible in the transaction snapshot.

20 years agoAdjust documentation examples to match recent error message rewordings.
Tom Lane [Tue, 30 Sep 2003 03:22:33 +0000 (03:22 +0000)]
Adjust documentation examples to match recent error message rewordings.
A couple other minor fixes.

20 years agoMinor copy-editing.
Tom Lane [Tue, 30 Sep 2003 01:56:11 +0000 (01:56 +0000)]
Minor copy-editing.

20 years agoUse slightly less broken markup.
Tom Lane [Tue, 30 Sep 2003 01:26:29 +0000 (01:26 +0000)]
Use slightly less broken markup.

20 years agoheap_open => relation_open to avoid unwanted restriction on relkind.
Tom Lane [Tue, 30 Sep 2003 00:59:51 +0000 (00:59 +0000)]
heap_open => relation_open to avoid unwanted restriction on relkind.
Per gripe from Gaetano Mendola.

20 years agoAdjust btree index build procedure so that the btree metapage looks
Tom Lane [Mon, 29 Sep 2003 23:40:26 +0000 (23:40 +0000)]
Adjust btree index build procedure so that the btree metapage looks
invalid (has the wrong magic number) until the build is entirely
complete.  This turns out to cost no additional writes in the normal
case, since we were rewriting the metapage at the end of the process
anyway.  In normal scenarios there's no real gain in security, because
a failed index build would roll back the transaction leaving an unused
index file, but for rebuilding shared system indexes this seems to add
some useful protection.

20 years agoRelease updates by Tom.
Bruce Momjian [Mon, 29 Sep 2003 23:16:44 +0000 (23:16 +0000)]
Release updates by Tom.

20 years agoImprove context display for failures during COPY IN, as recently
Tom Lane [Mon, 29 Sep 2003 22:06:40 +0000 (22:06 +0000)]
Improve context display for failures during COPY IN, as recently
discussed on pghackers.

20 years agoUpdate Jan's completed item.
Bruce Momjian [Mon, 29 Sep 2003 21:42:04 +0000 (21:42 +0000)]
Update Jan's completed item.

20 years agoRemove, already done:
Bruce Momjian [Mon, 29 Sep 2003 21:41:29 +0000 (21:41 +0000)]
Remove, already done:

<  o Fixed PL/Tcl's spi_prepare to accept full qualified type names in
<    the parameter type list

20 years agoUpdate HISTORY for 7.4.
Bruce Momjian [Mon, 29 Sep 2003 21:25:43 +0000 (21:25 +0000)]
Update HISTORY for 7.4.

Neil Conway

20 years agoAdd:
Bruce Momjian [Mon, 29 Sep 2003 21:15:19 +0000 (21:15 +0000)]
Add:

>  o Fixed PL/Tcl's spi_prepare to accept full qualified type names in
>    the parameter type list

20 years agoFixed type lookup in spi_prepare for possible qualified
Jan Wieck [Mon, 29 Sep 2003 19:24:06 +0000 (19:24 +0000)]
Fixed type lookup in spi_prepare for possible qualified
type name specification.

Jan

20 years agoThe brackets aren't put on the CHECK constraints properly.
Bruce Momjian [Mon, 29 Sep 2003 18:55:56 +0000 (18:55 +0000)]
The brackets aren't put on the CHECK constraints properly.

Before patch:

test=# select pg_get_constraintdef(oid) from pg_constraint;
                                       pg_get_constraintdef
-------------------------------------------------------------------------------------------------
  CHECK (VALUE >= 0)
  CHECK ((((a)::text = 'asdf'::text) OR ((a)::text = 'fdsa'::text)) OR
((a)::text = 'dfd'::text))
  PRIMARY KEY (b)
  FOREIGN KEY (a) REFERENCES test2(b)
  UNIQUE (b)
(5 rows)

test=# select pg_get_constraintdef(oid, true) from pg_constraint;
                                pg_get_constraintdef
-----------------------------------------------------------------------------------
  CHECK VALUE >= 0
  CHECK a::text = 'asdf'::text OR a::text = 'fdsa'::text OR a::text =
'dfd'::text
  PRIMARY KEY (b)
  FOREIGN KEY (a) REFERENCES test2(b)
  UNIQUE (b)
(5 rows)

After patch:

test=# select pg_get_constraintdef(oid) from pg_constraint;
                                       pg_get_constraintdef
-------------------------------------------------------------------------------------------------
  CHECK (VALUE >= 0)
  CHECK ((((a)::text = 'asdf'::text) OR ((a)::text = 'fdsa'::text)) OR
((a)::text = 'dfd'::text))
  PRIMARY KEY (b)
  FOREIGN KEY (a) REFERENCES test2(b)
  UNIQUE (b)
(5 rows)

test=# select pg_get_constraintdef(oid, true) from pg_constraint;
                                pg_get_constraintdef
-----------------------------------------------------------------------------------
  CHECK (VALUE >= 0)

`  CHECK (a::text = 'asdf'::text OR a::text = 'fdsa'::text OR a::text =
'dfd'::text)
  PRIMARY KEY (b)
  FOREIGN KEY (a) REFERENCES test2(b)
  UNIQUE (b)
(5 rows)

It's important that those brackets are there to (a) match all other
constraints and (b) so that people can just copy and paste them and it
will work as SQL.

Christopher Kings-Lynne

20 years agoI've not changed any malloc/calloc to palloc. It looks to me that these memory
Bruce Momjian [Mon, 29 Sep 2003 18:54:38 +0000 (18:54 +0000)]
I've not changed any malloc/calloc to palloc. It looks to me that these memory
areas are for the lifetime of the backend and in the interests of not breaking
something that's not broken I left alone.

Note for anyone reading this and wanting it for tsearch-v2-stable (i.e. for 7.3
backend) this patch probably will not apply cleanly to that source. It should
be simple enough to see what's going on and apply the changes by hand if need
be.

--
Nigel J. Andrews

20 years agoThis patch fixes an obvious bug in the "should I print the duration of
Bruce Momjian [Mon, 29 Sep 2003 18:50:58 +0000 (18:50 +0000)]
This patch fixes an obvious bug in the "should I print the duration of
this query?" logic in postgres.c

Also, make it print "duration:" like log_duration.

Neil Conway

20 years agoI discovered that TupleDescGetAttInMetadata and BuildTupleFromCStrings
Bruce Momjian [Mon, 29 Sep 2003 18:22:48 +0000 (18:22 +0000)]
I discovered that TupleDescGetAttInMetadata and BuildTupleFromCStrings
don't deal well with tuples having dropped columns. The attached fixes
the issue. Please apply.

Joe Conway

20 years ago> >
Bruce Momjian [Mon, 29 Sep 2003 18:21:33 +0000 (18:21 +0000)]
> >
> >  a) Write documentation how the win32 console needs to be set up so that
> >     psql can handle 8-bit characters.
> >     Where should it be added? The Section "Installation on Windows" in the
> >     Administrator's Guide seems natural to me.
> >
> >  b) Add code to psql that prints a warning on startup of psql when the
> >     console codepage differs from the windows codepage, something like
> >
> >     Warning: Console codepage (850) differs from windows codepage (1252)
> >              8-bit characters will not work correctly. See PostgreSQL
> >              documentation "Installation on Windows" for details.
>
Attached are two patches:

 - installdoc.patch contains an additional paragraph on the win32 console
   codepage for the chapter "Installation on Windows"
   Due to a lack of SGML-tools, I have only edited the text and not tested
   the SGML code - please check it before merging into the CVS branch.

 - psqlcodepage.patch adds the warning about a problematic codepage to psql.

Christoph Dalitz

20 years agoThis patch makes a few incremental improvements to geqo.sgml and
Bruce Momjian [Mon, 29 Sep 2003 18:18:35 +0000 (18:18 +0000)]
This patch makes a few incremental improvements to geqo.sgml and
arch-dev.sgml

Neil Conway

20 years agoA) Fixes a bug that prevented mirroring of data on a table that has had
Bruce Momjian [Mon, 29 Sep 2003 18:16:48 +0000 (18:16 +0000)]
A) Fixes a bug that prevented mirroring of data on a table that has had
a column dropped.

B) Updated the documentation for the 7.4 release.

Steven Singer

20 years agoChange 7.3.4 release note format to match the rest of the 7.3.x
Bruce Momjian [Mon, 29 Sep 2003 18:15:12 +0000 (18:15 +0000)]
Change 7.3.4 release note format to match the rest of the 7.3.x
releases.

Rod Taylor

20 years agoUpdate release notes.
Bruce Momjian [Mon, 29 Sep 2003 18:06:48 +0000 (18:06 +0000)]
Update release notes.

20 years agoAdd:
Bruce Momjian [Mon, 29 Sep 2003 16:52:14 +0000 (16:52 +0000)]
Add:

> * Have SHOW ALL and pg_settings show descriptions for server-side variables(Joe)

20 years agoAdjust the new Norwegian translation for some of the easier message
Peter Eisentraut [Mon, 29 Sep 2003 16:41:33 +0000 (16:41 +0000)]
Adjust the new Norwegian translation for some of the easier message
changes between 7.3 and 7.4, for example quoting and function names.

20 years agoApparently, gettext doesn't like double parentheses around argument. What
Peter Eisentraut [Mon, 29 Sep 2003 16:39:18 +0000 (16:39 +0000)]
Apparently, gettext doesn't like double parentheses around argument.  What
were they doing here anyway?

20 years agoMake message fit guidelines.
Peter Eisentraut [Mon, 29 Sep 2003 16:38:04 +0000 (16:38 +0000)]
Make message fit guidelines.

20 years agoEliminate another gratuitous message wording difference.
Peter Eisentraut [Mon, 29 Sep 2003 16:37:29 +0000 (16:37 +0000)]
Eliminate another gratuitous message wording difference.

20 years agoNew Norwegian translation by Trond Endrestøl, actually made for 7.3, but
Peter Eisentraut [Mon, 29 Sep 2003 10:57:06 +0000 (10:57 +0000)]
New Norwegian translation by Trond Endrestøl, actually made for 7.3, but
this should help people get started in 7.4 as well.

20 years agoNew translations
Peter Eisentraut [Mon, 29 Sep 2003 09:51:29 +0000 (09:51 +0000)]
New translations

20 years agoUpdate HISTORY to current.
Bruce Momjian [Mon, 29 Sep 2003 04:50:36 +0000 (04:50 +0000)]
Update HISTORY to current.

20 years agoFix #error message to mention renamed option --disable-spinlocks.
Bruce Momjian [Mon, 29 Sep 2003 04:20:22 +0000 (04:20 +0000)]
Fix #error message to mention renamed option --disable-spinlocks.

20 years agoUpdate docs that point to thread test program.
Bruce Momjian [Mon, 29 Sep 2003 03:32:05 +0000 (03:32 +0000)]
Update docs that point to thread test program.

20 years agoFix broken definition of :print: character class, per Bruno Wolff.
Tom Lane [Mon, 29 Sep 2003 00:21:58 +0000 (00:21 +0000)]
Fix broken definition of :print: character class, per Bruno Wolff.
Also, make :alnum: character class directly dependent on isalnum()
rather than guessing.

20 years agoTranslation update
Peter Eisentraut [Mon, 29 Sep 2003 00:17:42 +0000 (00:17 +0000)]
Translation update

20 years agoMore message editing, some suggested by Alvaro Herrera
Peter Eisentraut [Mon, 29 Sep 2003 00:05:25 +0000 (00:05 +0000)]
More message editing, some suggested by Alvaro Herrera

20 years agoRestructure plpgsql's caching of 'simple' expression evaluation trees
Tom Lane [Sun, 28 Sep 2003 23:37:45 +0000 (23:37 +0000)]
Restructure plpgsql's caching of 'simple' expression evaluation trees
to be less dangerous, and often faster as well.  ExprState trees are
not kept across transaction boundaries; this eliminates problems with
resource leakage in failed transactions.  But by keeping them in a
per-transaction EState, we can safely arrange for a single ExprState
to be shared by all the expression evaluations done in a given plpgsql
function call.  (Formerly it seemed necessary to create and destroy an
ExprState for each exec_eval_simple_expr() call.)  This saves time in
any scenario where a plpgsql function executes more than one expression.
Seems to be about as fast as 7.3 for simple cases, and significantly
faster for functions that do a lot of calculations.

20 years agoAdd a mechanism to let dynamically loaded modules register post-commit/
Tom Lane [Sun, 28 Sep 2003 23:26:20 +0000 (23:26 +0000)]
Add a mechanism to let dynamically loaded modules register post-commit/
post-abort cleanup hooks.  I'm surprised that we have not needed this
already, but I need it now to fix a plpgsql problem, and the usefulness
for other dynamically loaded modules seems obvious.

20 years agoNow that we have UPDATE tab SET col = DEFAULT, get rid of horrid hack
Tom Lane [Sun, 28 Sep 2003 02:11:23 +0000 (02:11 +0000)]
Now that we have UPDATE tab SET col = DEFAULT, get rid of horrid hack
in the RI triggers for ON DELETE/UPDATE SET DEFAULT.  The code depended
way too much on knowledge of plan structure, and yet still would fail
if the generated query got rewritten by rules.

20 years agoUpdate docs on function call permissions in view, from Tom.
Bruce Momjian [Sun, 28 Sep 2003 01:19:33 +0000 (01:19 +0000)]
Update docs on function call permissions in view, from Tom.

20 years agoAdjust pgindent for newer awks.
Bruce Momjian [Sun, 28 Sep 2003 00:25:22 +0000 (00:25 +0000)]
Adjust pgindent for newer awks.

Nigel J. Andrews

20 years agoAllow pgindent to work with newer BSD indents.
Bruce Momjian [Sun, 28 Sep 2003 00:22:58 +0000 (00:22 +0000)]
Allow pgindent to work with newer BSD indents.

20 years agoMake dbf2pg safe for non-ASCII character sets.
Bruce Momjian [Sat, 27 Sep 2003 22:24:49 +0000 (22:24 +0000)]
Make dbf2pg safe for non-ASCII character sets.

M?rcio Dick Smiderle

20 years agoMark Linux for threads.
Bruce Momjian [Sat, 27 Sep 2003 22:23:35 +0000 (22:23 +0000)]
Mark Linux for threads.

20 years agoRemove erroneous restriction that -t cannot be used to select a
Tom Lane [Sat, 27 Sep 2003 22:10:01 +0000 (22:10 +0000)]
Remove erroneous restriction that -t cannot be used to select a
sequence for dumping.

20 years agoUpdate bsd indent patch.
Bruce Momjian [Sat, 27 Sep 2003 21:26:09 +0000 (21:26 +0000)]
Update bsd indent patch.

20 years agoCleanup pgindent patch.
Bruce Momjian [Sat, 27 Sep 2003 21:19:47 +0000 (21:19 +0000)]
Cleanup pgindent patch.

20 years agoGot the link order wrong :-(
Tom Lane [Sat, 27 Sep 2003 20:38:19 +0000 (20:38 +0000)]
Got the link order wrong :-(

20 years agoAdjust Darwin build to use the default 'two level namespace' linking
Tom Lane [Sat, 27 Sep 2003 19:35:32 +0000 (19:35 +0000)]
Adjust Darwin build to use the default 'two level namespace' linking
method.  Fix a number of places where shared libraries were linked without
mentioning all the libraries they depend on; the Darwin and AIX ports
are known to require this, and it doesn't seem to hurt any other supported
platforms.  (Hence, remove code in pl/tcl makefile that tried to avoid
mentioning other libs if not needed.)

20 years agoFixed calculation of bid when generating accounts. Used to create
Jan Wieck [Sat, 27 Sep 2003 19:15:34 +0000 (19:15 +0000)]
Fixed calculation of bid when generating accounts. Used to create
accounts.bid values of zero.

Jan

20 years agoFix typo in message.
Tom Lane [Sat, 27 Sep 2003 18:16:35 +0000 (18:16 +0000)]
Fix typo in message.

20 years agoAdd compile step to instructions.
Bruce Momjian [Sat, 27 Sep 2003 17:29:30 +0000 (17:29 +0000)]
Add compile step to instructions.

20 years agoMake dump files created by initdb have consistent EOL termination, to
Bruce Momjian [Sat, 27 Sep 2003 16:27:57 +0000 (16:27 +0000)]
Make dump files created by initdb have consistent EOL termination, to
pass COPY's EOL tests.

20 years agoFinalize configuration of thread_test program.
Bruce Momjian [Sat, 27 Sep 2003 16:24:45 +0000 (16:24 +0000)]
Finalize configuration of thread_test program.

20 years agoChanged the logic when a CAST is dumped according to discussion
Jan Wieck [Sat, 27 Sep 2003 15:34:06 +0000 (15:34 +0000)]
Changed the logic when a CAST is dumped according to discussion
on pgsql-hackers.

A cast is included in the dump output if any of the objects does
not belong to a system namespace and all of the non-system namespace
objects belong to dumped namespaces. System namespace is defined
as nspname begins with "pg_".

Jan

20 years agoRename thread compile flag. Move thread test program to tools/thread,
Bruce Momjian [Sat, 27 Sep 2003 15:32:48 +0000 (15:32 +0000)]
Rename thread compile flag.  Move thread test program to tools/thread,
and improve tests.

20 years agoSmall SGML cleanup from Jeroen Ruigrok/asmodai
Bruce Momjian [Sat, 27 Sep 2003 14:06:21 +0000 (14:06 +0000)]
Small SGML cleanup from Jeroen Ruigrok/asmodai

20 years agoYou can't NLS-enable a program component by just putting gettext() around
Peter Eisentraut [Sat, 27 Sep 2003 09:29:31 +0000 (09:29 +0000)]
You can't NLS-enable a program component by just putting gettext() around
every string, especially if some of the output should be fixed-format
machine-readable.  This needs to be more carefully sorted out. Also, make
the help message generated by --help-config -h be more similar in style to
the others.

20 years agoThe formatting of the display of the locale names assumed that locale names
Peter Eisentraut [Sat, 27 Sep 2003 09:21:26 +0000 (09:21 +0000)]
The formatting of the display of the locale names assumed that locale names
are not longer than 8 characters.  But sometimes they are, and that made
the display quite ugly.  So just format them vertically so that everyone
can read them.

20 years agoDocument that functions are checked independently of the view
Bruce Momjian [Sat, 27 Sep 2003 00:10:31 +0000 (00:10 +0000)]
Document that functions are checked independently of the view
permissions.

20 years agoAdd "tuning" section to install instructions.
Bruce Momjian [Fri, 26 Sep 2003 17:47:44 +0000 (17:47 +0000)]
Add "tuning" section to install instructions.

20 years agoMove -D_GNU_SOURCE hack from port header to template, so that
Tom Lane [Fri, 26 Sep 2003 17:39:13 +0000 (17:39 +0000)]
Move -D_GNU_SOURCE hack from port header to template, so that
configure's tests see the same compilation environment as the code.
Per discussion with Stephan Szabo.

20 years agoVarious message fixes, among those fixes for the previous round of fixes
Peter Eisentraut [Fri, 26 Sep 2003 15:27:37 +0000 (15:27 +0000)]
Various message fixes, among those fixes for the previous round of fixes

20 years agoTranslation update
Peter Eisentraut [Fri, 26 Sep 2003 15:25:39 +0000 (15:25 +0000)]
Translation update

20 years agoFixed segfault after error in parsing precision argument.
Michael Meskes [Fri, 26 Sep 2003 15:16:29 +0000 (15:16 +0000)]
Fixed segfault after error in parsing precision argument.

20 years agoFix tid scan evaluation of non-constant TID values; can't try to do it
Tom Lane [Fri, 26 Sep 2003 01:17:01 +0000 (01:17 +0000)]
Fix tid scan evaluation of non-constant TID values; can't try to do it
during ExecInitTidScan, because the rest of the executor isn't ready.

20 years agoMake the world safe (more or less) for dropped columns in plpgsql rowtypes.
Tom Lane [Thu, 25 Sep 2003 23:02:12 +0000 (23:02 +0000)]
Make the world safe (more or less) for dropped columns in plpgsql rowtypes.

20 years agotlist_matches_tupdesc() needs to defend itself against dropped columns.
Tom Lane [Thu, 25 Sep 2003 19:41:49 +0000 (19:41 +0000)]
tlist_matches_tupdesc() needs to defend itself against dropped columns.

20 years agoGet rid of ReferentialIntegritySnapshotOverride by extending Executor API
Tom Lane [Thu, 25 Sep 2003 18:58:36 +0000 (18:58 +0000)]
Get rid of ReferentialIntegritySnapshotOverride by extending Executor API
to allow es_snapshot to be set to SnapshotNow rather than a query snapshot.
This solves a bug reported by Wade Klaver, wherein triggers fired as a
result of RI cascade updates could misbehave.

20 years agoFix regression error messages for platforms Peter doesn't use.
Tom Lane [Thu, 25 Sep 2003 16:16:35 +0000 (16:16 +0000)]
Fix regression error messages for platforms Peter doesn't use.

20 years agoFix grammatical error introduced into error message.
Tom Lane [Thu, 25 Sep 2003 15:58:06 +0000 (15:58 +0000)]
Fix grammatical error introduced into error message.

20 years agoForget to update the real source of the regression test results after
Peter Eisentraut [Thu, 25 Sep 2003 15:48:17 +0000 (15:48 +0000)]
Forget to update the real source of the regression test results after
message editing.

20 years agoMessage editing: remove gratuitous variations in message wording, standardize
Peter Eisentraut [Thu, 25 Sep 2003 06:58:07 +0000 (06:58 +0000)]
Message editing: remove gratuitous variations in message wording, standardize
terms, add some clarifications, fix some untranslatable attempts at dynamic
message building.

20 years agoUpdate Japanese FAQ, from Jun Kuwamura
Bruce Momjian [Thu, 25 Sep 2003 02:27:54 +0000 (02:27 +0000)]
Update Japanese FAQ, from Jun Kuwamura

20 years agoFix creation of temporary sequences in ecpg, per report from Edmund Bacon.
Bruce Momjian [Wed, 24 Sep 2003 19:05:32 +0000 (19:05 +0000)]
Fix creation of temporary sequences in ecpg, per report from Edmund Bacon.

20 years agoRepair some REINDEX problems per recent discussions. The relcache is
Tom Lane [Wed, 24 Sep 2003 18:54:02 +0000 (18:54 +0000)]
Repair some REINDEX problems per recent discussions.  The relcache is
now able to cope with assigning new relfilenode values to nailed-in-cache
indexes, so they can be reindexed using the fully crash-safe method.  This
leaves only shared system indexes as special cases.  Remove the 'index
deactivation' code, since it provides no useful protection in the shared-
index case.  Require reindexing of shared indexes to be done in standalone
mode, but remove other restrictions on REINDEX.  -P (IgnoreSystemIndexes)
now prevents using indexes for lookups, but does not disable index updates.
It is therefore safe to allow from PGOPTIONS.  Upshot: reindexing system catalogs
can be done without a standalone backend for all cases except
shared catalogs.

20 years agoanother totally benign, test commit
Marc G. Fournier [Wed, 24 Sep 2003 18:09:15 +0000 (18:09 +0000)]
another totally benign, test commit

20 years agotest 2, benign, no change
Marc G. Fournier [Wed, 24 Sep 2003 18:08:20 +0000 (18:08 +0000)]
test 2, benign, no change

20 years agonot change, just added a space to test something
Marc G. Fournier [Wed, 24 Sep 2003 18:05:36 +0000 (18:05 +0000)]
not change, just added a space to test something

20 years agoAdd positive defense against dropped columns, per suggestion from
Tom Lane [Wed, 24 Sep 2003 05:38:38 +0000 (05:38 +0000)]
Add positive defense against dropped columns, per suggestion from
Christopher Kings-Lynne.  Also fix a couple more schema-awareness issues.

20 years agoIssue CREATE SCHEMA as the originally connecting user, with an
Tom Lane [Tue, 23 Sep 2003 23:31:52 +0000 (23:31 +0000)]
Issue CREATE SCHEMA as the originally connecting user, with an
AUTHORIZATION clause to specify the desired owner.  This allows a
superuser to restore schemas owned by users without CREATE-SCHEMA
permissions (ie, schemas originally created by a superuser using
AUTHORIZATION).  --no-owner can be specified to suppress the
AUTHORIZATION clause if need be.

20 years agopg_dump/pg_restore now always use SET SESSION AUTHORIZATION, not \connect,
Tom Lane [Tue, 23 Sep 2003 22:48:53 +0000 (22:48 +0000)]
pg_dump/pg_restore now always use SET SESSION AUTHORIZATION, not \connect,
to control object ownership.  The use-set-session-authorization and
no-reconnect switches are obsolete (still accepted on the command line,
but they don't do anything).  This is a precursor to fixing handling
of CREATE SCHEMA, which will be a separate commit.

20 years agoAdd:
Bruce Momjian [Tue, 23 Sep 2003 21:00:18 +0000 (21:00 +0000)]
Add:

> * Allow LISTEN/NOTIFY to store info in memory rather than tables

20 years agoAdd:
Bruce Momjian [Tue, 23 Sep 2003 20:45:58 +0000 (20:45 +0000)]
Add:

* Allow major upgrades without dump/reload, perhaps using pg_upgrade

20 years agoMinor copy-editing for plpgsql chapter.
Tom Lane [Tue, 23 Sep 2003 19:58:50 +0000 (19:58 +0000)]
Minor copy-editing for plpgsql chapter.

20 years agoTweak generic_type_consistency routines to avoid loss of functionality
Tom Lane [Tue, 23 Sep 2003 17:12:53 +0000 (17:12 +0000)]
Tweak generic_type_consistency routines to avoid loss of functionality
since 7.3: 'select array_dims(histogram_bounds) from pg_stats' used to
work and still should.  Problem was that code wouldn't take input of
declared type anyarray as matching an anyarray argument.  Allow this
case as long as we don't need to determine an element type (which in
practice means as long as anyelement isn't used in the function signature).

20 years ago_SPI_cursor_operation forgot to check for failure return from
Tom Lane [Tue, 23 Sep 2003 15:11:33 +0000 (15:11 +0000)]
_SPI_cursor_operation forgot to check for failure return from
_SPI_begin_call.  Per gripe from Tomasz Myrta.

20 years ago- Changed struct definition handling so "struct foo {}" always gets defined.
Michael Meskes [Tue, 23 Sep 2003 12:56:35 +0000 (12:56 +0000)]
- Changed struct definition handling so "struct foo {}" always gets defined.

20 years agoFix to properly handle timezone offsets that are partial hours. If the offset
Barry Lind [Tue, 23 Sep 2003 06:13:52 +0000 (06:13 +0000)]
Fix to properly handle timezone offsets that are partial hours.  If the offset
was a partial hour and less than gmt (i.e. -02:30) the code would corrupt the
minutes part.

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java

20 years agoA fix for jdbc regression test submitted yesterday by Kim Ho at Redhat
Barry Lind [Tue, 23 Sep 2003 06:11:06 +0000 (06:11 +0000)]
A fix for jdbc regression test submitted yesterday by Kim Ho at Redhat

 Modified Files:
  ResultSetTest.java

20 years agoPutting back the previous change must be the first thing.
Hiroshi Inoue [Tue, 23 Sep 2003 01:51:09 +0000 (01:51 +0000)]
Putting back the previous change must be the first thing.
ALso put back a #ifndef ENABLE_REINDEX_NAILED_RELATIONS
which was removed about a year ago.

20 years agofix hlfinditem function. Thanks to "Stphane Bidoul" <stephane.bidoul@softwareag.com>.
Teodor Sigaev [Mon, 22 Sep 2003 13:32:33 +0000 (13:32 +0000)]
fix hlfinditem function. Thanks to "Stphane Bidoul" <stephane.bidoul@softwareag.com>.

The 'word' variable there is initialised from
the prs->words array, but immediately after,
that array may be reallocated, thus leaving
word pointing to unallocated memory.

20 years ago- Fixed order mismatch in processing "using" arguments.
Michael Meskes [Mon, 22 Sep 2003 13:19:39 +0000 (13:19 +0000)]
- Fixed order mismatch in processing "using" arguments.
- Fixed some minor things in test cases.
- Use defines for Informix error codes.

20 years agoAdditional jdbc regression tests submitted by Oliver Jowett. Some tests are
Barry Lind [Mon, 22 Sep 2003 05:38:01 +0000 (05:38 +0000)]
Additional jdbc regression tests submitted by Oliver Jowett.  Some tests are
currently commented out, pending fixes for the bugs these tests uncovered.

 Modified Files:
  jdbc/org/postgresql/test/jdbc2/Jdbc2TestSuite.java
  jdbc/org/postgresql/test/jdbc2/ServerPreparedStmtTest.java
 Added Files:
  jdbc/org/postgresql/test/jdbc2/CursorFetchTest.java

20 years agoApplied jdbc patch from Kim Ho at RedHat to improve boolean support, as well
Barry Lind [Mon, 22 Sep 2003 04:55:00 +0000 (04:55 +0000)]
Applied jdbc patch from Kim Ho at RedHat to improve boolean support, as well
as some additional regression tests for this an other recent changes.

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1ResultSet.java
  jdbc/org/postgresql/test/jdbc2/DateTest.java
  jdbc/org/postgresql/test/jdbc2/ResultSetTest.java
  jdbc/org/postgresql/test/jdbc2/TimeTest.java
  jdbc/org/postgresql/test/jdbc2/TimestampTest.java

20 years agoHeapTupleSatisfiesVacuum() needs to be more careful about the
Tom Lane [Mon, 22 Sep 2003 00:47:23 +0000 (00:47 +0000)]
HeapTupleSatisfiesVacuum() needs to be more careful about the
difference between INSERT_IN_PROGRESS and DELETE_IN_PROGRESS for
tuples inserted and then deleted by a concurrent transaction.
Example of bug:
regression=# create table foo (f1 int);
CREATE TABLE
regression=# begin;
BEGIN
regression=# insert into foo values(1);
INSERT 195531 1
regression=# delete from foo;
DELETE 1
regression=# insert into foo values(1);
INSERT 195532 1
regression=# create unique index fooi on foo(f1);
ERROR:  could not create unique index
DETAIL:  Table contains duplicated values.

20 years agoTranslation updates
Peter Eisentraut [Mon, 22 Sep 2003 00:27:01 +0000 (00:27 +0000)]
Translation updates

20 years agoMessage wording improvements
Peter Eisentraut [Mon, 22 Sep 2003 00:23:35 +0000 (00:23 +0000)]
Message wording improvements

20 years agoMake the SQL command synopses appear less random.
Peter Eisentraut [Mon, 22 Sep 2003 00:16:58 +0000 (00:16 +0000)]
Make the SQL command synopses appear less random.

20 years agoArrange to align shared disk buffers on at least 32-byte boundaries,
Tom Lane [Sun, 21 Sep 2003 17:57:21 +0000 (17:57 +0000)]
Arrange to align shared disk buffers on at least 32-byte boundaries,
not just MAXALIGN boundaries.  This makes a noticeable difference in
the speed of transfers to and from kernel space, at least on recent
Pentiums, and might help other CPUs too.  We should look at making
this happen for local buffers and buffile.c too.  Patch from Manfred Spraul.

20 years agoFix some ill-considered ifdefs, per report from Hans-JÏrgen SchÎnig.
Tom Lane [Sun, 21 Sep 2003 17:42:22 +0000 (17:42 +0000)]
Fix some ill-considered ifdefs, per report from Hans-JÏrgen SchÎnig.