]> granicus.if.org Git - postgresql/log
postgresql
21 years agoAvoid generating invalid character encoding sequences in make_greater_string.
Tom Lane [Mon, 2 Feb 2004 03:07:25 +0000 (03:07 +0000)]
Avoid generating invalid character encoding sequences in make_greater_string.
Not sure how this mistake evaded detection for so long.

21 years agoFix text_position to not scan past end of source string in multibyte
Joe Conway [Sun, 1 Feb 2004 04:05:13 +0000 (04:05 +0000)]
Fix text_position to not scan past end of source string in multibyte
case, per report from Korea PostgreSQL Users' Group. Copied from Tom
Lane's 7.4 branch patch.

21 years agoRepair incorrect order of operations in GetNewTransactionId(). We must
Tom Lane [Mon, 26 Jan 2004 19:16:40 +0000 (19:16 +0000)]
Repair incorrect order of operations in GetNewTransactionId().  We must
complete ExtendCLOG() before advancing nextXid, so that if that routine
fails, the next incoming transaction will try it again.  Per trouble
report from Christopher Kings-Lynne.

21 years agoRevert ill-starred change of 13-Feb-02: it appeared to fix a problem of
Tom Lane [Wed, 14 Jan 2004 03:39:36 +0000 (03:39 +0000)]
Revert ill-starred change of 13-Feb-02: it appeared to fix a problem of
incorrect permissions checking, but in fact disabled most all permissions
checks for view updates.  This corrects problems reported by Sergey
Yatskevich among others, at the cost of re-introducing the problem
previously reported by Tim Burgess.  However, since we'd lived with that
problem for quite awhile without knowing it, we can live with it awhile
longer until a proper fix can be made in 7.5.

21 years agoRepair indexed bytea like operations, and related selectivity
Joe Conway [Sun, 7 Dec 2003 04:11:26 +0000 (04:11 +0000)]
Repair indexed bytea like operations, and related selectivity
functionality. Per bug report by Alvar Freude:
http://archives.postgresql.org/pgsql-bugs/2003-12/msg00022.php

21 years agoBrand 7.3.5. REL7_3_5
Tom Lane [Tue, 2 Dec 2003 16:26:00 +0000 (16:26 +0000)]
Brand 7.3.5.

21 years agoUpdate release notes for 7.3.5.
Tom Lane [Tue, 2 Dec 2003 16:15:47 +0000 (16:15 +0000)]
Update release notes for 7.3.5.

21 years agoForce zero_damaged_pages to be effectively ON during recovery from WAL,
Tom Lane [Mon, 1 Dec 2003 16:53:42 +0000 (16:53 +0000)]
Force zero_damaged_pages to be effectively ON during recovery from WAL,
since there is no need to worry about damaged pages when we are going to
overwrite them anyway from the WAL.  Per recent discussion.

21 years agoBack-patch fix to check vartypmod when matching PlannerParamVar entries.
Tom Lane [Sun, 30 Nov 2003 22:24:16 +0000 (22:24 +0000)]
Back-patch fix to check vartypmod when matching PlannerParamVar entries.
This should prevent some obscure cases of 'variable not in subplan target
lists', although actual failures have only been reported against 7.4 in
which the bug is much easier to trigger.

21 years agoBack-patch fix to cause stats processes to detach from shared memory,
Tom Lane [Sun, 30 Nov 2003 21:56:36 +0000 (21:56 +0000)]
Back-patch fix to cause stats processes to detach from shared memory,
so that they do not prevent the postmaster from deleting the shmem segment
during crash recovery.

21 years agoMake PQescapeBytea and byteaout consistent with each other, and
Joe Conway [Sun, 30 Nov 2003 20:52:37 +0000 (20:52 +0000)]
Make PQescapeBytea and byteaout consistent with each other, and
octal escape all octets outside the range 0x20 to 0x7e. This fixes
the problem pointed out by Sergey Yatskevich here:
http://archives.postgresql.org/pgsql-bugs/2003-11/msg00140.php

21 years agoAdded missing SPI_finish() calls to get_tuple_of_interest(). Fixes bug
Joe Conway [Wed, 26 Nov 2003 20:44:54 +0000 (20:44 +0000)]
Added missing SPI_finish() calls to get_tuple_of_interest(). Fixes bug
reported by Andrea Grassi.

21 years agoBack-patch removal of unnecessary inclusion of <crypt.h>, so that
Tom Lane [Fri, 31 Oct 2003 18:14:11 +0000 (18:14 +0000)]
Back-patch removal of unnecessary inclusion of <crypt.h>, so that
7.3.5 will build out-of-the-box on Solaris.

21 years agoMoved the recently added test for foreign key disabled by rewrite
Jan Wieck [Fri, 31 Oct 2003 14:27:51 +0000 (14:27 +0000)]
Moved the recently added test for foreign key disabled by rewrite
rule into the rule.sql since it affects the latter if run in paralell.

Jan

21 years agoFix for possible referential integrity violation when a qualified ON INSERT
Jan Wieck [Fri, 31 Oct 2003 03:57:42 +0000 (03:57 +0000)]
Fix for possible referential integrity violation when a qualified ON INSERT
rule split the query into one INSERT and one UPDATE where the UPDATE
then hit's the just created row without modifying the key fields again.
In this special case, the new key slipped in totally unchecked.

Jan

21 years agoSupport for qualified type names in PL/Tcl's spi_prepare command.
Jan Wieck [Thu, 30 Oct 2003 02:00:44 +0000 (02:00 +0000)]
Support for qualified type names in PL/Tcl's spi_prepare command.

This is not 100% backward compatible as formerly a double quoted
type name containing a dot could be used. But I don't think may people
use dot's in the name of user defined types.

Jan

21 years agoDon't choke when the handler for a procedural language is located in
Tom Lane [Tue, 28 Oct 2003 21:05:39 +0000 (21:05 +0000)]
Don't choke when the handler for a procedural language is located in
the pg_catalog schema.  Per bug report some months back from Jochem van Dieten.

21 years agoIt is possible for ResolveNew to be used to insert a sublink into a
Tom Lane [Mon, 20 Oct 2003 20:02:30 +0000 (20:02 +0000)]
It is possible for ResolveNew to be used to insert a sublink into a
subquery that didn't previously have one.  We have traditionally made
the caller of ResolveNew responsible for updating the hasSubLinks flag
of the outermost query, but this fails to account for hasSubLinks in
subqueries.  Fix ResolveNew to handle this.  We might later want to
change the calling convention of ResolveNew so that it can fix the
outer query too, simplifying callers.  But I went with the localized
fix for now.  Per bug report from J Smith, 20-Oct-03.

21 years agoWhen dumping CREATE INDEX, must show opclass name if the opclass isn't
Tom Lane [Thu, 2 Oct 2003 22:25:08 +0000 (22:25 +0000)]
When dumping CREATE INDEX, must show opclass name if the opclass isn't
in the schema search path.  Otherwise pg_dump doesn't correctly dump
scenarios where a custom opclass is created in 'public' and then used
by indexes in other schemas.

21 years ago[ Patch applied only to 7.3.X.]
Bruce Momjian [Mon, 29 Sep 2003 18:53:08 +0000 (18:53 +0000)]
[ Patch applied only to 7.3.X.]

Hi There's a bug in the clusterdb script where it looks like the arguments
to the psql command are being passed in the wrong order, so it fails when
you run it on a database that is not on localhost.
Here's the output from the command:

133 anands-Computer:bin/scripts> clusterdb -h wooster -U rr granada
psql: warning: extra option wooster ignored
psql: warning: extra option -U ignored
psql: warning: extra option rr ignored
psql: warning: extra option -F: ignored
psql: warning: extra option -P ignored
psql: warning: extra option format=unaligned ignored
psql: warning: extra option -t ignored
psql: warning: extra option -c ignored
psql: warning: extra option SELECT nspname, pg_class.relname,
pg_class_2.relname FROM pg_class, pg_class AS pg_class_2 JOIN pg_namespace
ON (pg_namespace.oid=relnamespace), pg_index WHERE
pg_class.oid=pg_index.indrelid AND pg_class_2.oid=pg_index.indexrelid AND
pg_index.indisclustered AND pg_class.relowner=(SELECT usesysid FROM
pg_user WHERE usename=current_user) ignored
psql: FATAL:  user "-h" does not exist

I'm attaching a patch that fixes the problem. The diff was run on
postgresql 7.3.4

Thanks a lot.
Anand Ranganathan

21 years agoBackpatched changes for rules when casts are dumped according to
Jan Wieck [Sun, 28 Sep 2003 17:46:19 +0000 (17:46 +0000)]
Backpatched changes for rules when casts are dumped according to
discussion on hackers.

Jan

21 years ago_SPI_cursor_operation forgot to check for failure return from
Tom Lane [Tue, 23 Sep 2003 15:11:41 +0000 (15:11 +0000)]
_SPI_cursor_operation forgot to check for failure return from
_SPI_begin_call.  Per gripe from Tomasz Myrta.

21 years agoBack-patch fix for plpython problems with dropped table columns;
Tom Lane [Wed, 17 Sep 2003 18:40:11 +0000 (18:40 +0000)]
Back-patch fix for plpython problems with dropped table columns;
per bug report from Arthur Ward, who also tested this patch.

21 years agoBack-patch the other part of Karel's formatting bug fix.
Tom Lane [Wed, 3 Sep 2003 19:01:35 +0000 (19:01 +0000)]
Back-patch the other part of Karel's formatting bug fix.

21 years agoRepair problems with to_char() overrunning its input string.
Tom Lane [Wed, 3 Sep 2003 15:00:07 +0000 (15:00 +0000)]
Repair problems with to_char() overrunning its input string.
From Karel Zak.

21 years agoFix translation mistake.
Peter Eisentraut [Sun, 24 Aug 2003 21:26:54 +0000 (21:26 +0000)]
Fix translation mistake.

21 years agoFix GB18030 to UTF-8 mapping table
Tatsuo Ishii [Sun, 24 Aug 2003 05:13:40 +0000 (05:13 +0000)]
Fix GB18030 to UTF-8 mapping table

21 years agoFix bug in GB18030 conversion script
Tatsuo Ishii [Sun, 24 Aug 2003 05:00:45 +0000 (05:00 +0000)]
Fix bug in GB18030 conversion script

21 years agoSigh, I'm an idiot ... SSL_ERROR_WANT_READ isn't an error condition
Tom Lane [Fri, 22 Aug 2003 21:57:44 +0000 (21:57 +0000)]
Sigh, I'm an idiot ... SSL_ERROR_WANT_READ isn't an error condition
at all, it just means 'no data available yet'.

21 years agoConversion functions must be STRICT to prevent them from getting null inputs.
Tom Lane [Fri, 8 Aug 2003 15:49:10 +0000 (15:49 +0000)]
Conversion functions must be STRICT to prevent them from getting null inputs.

21 years agoBackport to 7.3. Third try to fix the sql injection
Barry Lind [Thu, 7 Aug 2003 17:56:27 +0000 (17:56 +0000)]
Backport to 7.3.  Third try to fix the sql injection
vulnerability.  This fix completely removes the ability (hack) of being able
to bind a list of values in an in clause.  It was demonstrated that by allowing
that functionality you open up the possibility for certain types of
sql injection attacks.  The previous fix attempts all focused on preventing
the insertion of additional sql statements (the semi-colon problem:
xxx; any new sql statement here).  But that still left the ability to
change the where clause on the current statement or perform a subselect
which can circumvent applicaiton security logic and/or allow you to call
any stored function.

 Modified Files:
  Tag: REL7_3_STABLE
jdbc/org/postgresql/Driver.java.in
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java

21 years agoFix several places where fractional-second inputs were misprocessed
Tom Lane [Tue, 5 Aug 2003 17:39:36 +0000 (17:39 +0000)]
Fix several places where fractional-second inputs were misprocessed
in HAVE_INT64_TIMESTAMP cases, including two potential stack smashes
when more than six fractional digits were supplied.  Per bug report
from Philipp Reisner.

21 years agoSSL_read/SSL_write do not approximate the return conventions of recv()
Tom Lane [Mon, 4 Aug 2003 17:58:25 +0000 (17:58 +0000)]
SSL_read/SSL_write do not approximate the return conventions of recv()
and send() very well at all; and in any case we can't use retval==0
for EOF due to race conditions.  Make the same fixes in the backend as
are required in libpq.

21 years agoFix some more problems with testing error returns from SSL.
Tom Lane [Mon, 4 Aug 2003 17:25:29 +0000 (17:25 +0000)]
Fix some more problems with testing error returns from SSL.

21 years agoFix longstanding error in _bt_search(): should moveright at top of loop not
Tom Lane [Tue, 29 Jul 2003 22:18:48 +0000 (22:18 +0000)]
Fix longstanding error in _bt_search(): should moveright at top of loop not
bottom.  Otherwise we fail to moveright when the root page was split while
we were "in flight" to it.  This is not a significant problem when the root
is above the leaf level, but if the root was also a leaf (ie, a single-page
index just got split) we may return the wrong leaf page to the caller,
resulting in failure to find a key that is in fact present.  Bug has existed
at least since 7.1, probably forever.

21 years agoFix timestamp_date for HAVE_INT64_TIMESTAMP case. REL7_3_4
Tom Lane [Thu, 24 Jul 2003 04:38:27 +0000 (04:38 +0000)]
Fix timestamp_date for HAVE_INT64_TIMESTAMP case.

21 years agoAdd Hungarian/Turkish FAQ's to 7.3.4.
Bruce Momjian [Thu, 24 Jul 2003 00:53:57 +0000 (00:53 +0000)]
Add Hungarian/Turkish FAQ's to 7.3.4.

21 years agoAdd to 7.3.4 release notes:
Bruce Momjian [Thu, 24 Jul 2003 00:44:24 +0000 (00:44 +0000)]
Add to 7.3.4 release notes:

* Repair breakage in timestamp-to-date conversion for dates before 2000

21 years agoRepair 7.3 breakage in timestamp-to-date conversion for dates before 2000.
Tom Lane [Thu, 24 Jul 2003 00:21:31 +0000 (00:21 +0000)]
Repair 7.3 breakage in timestamp-to-date conversion for dates before 2000.

21 years agoPatch to fix additional SQL injection vulnerabilities reported by Oliver Jowett
Barry Lind [Wed, 23 Jul 2003 23:34:31 +0000 (23:34 +0000)]
Patch to fix additional SQL injection vulnerabilities reported by Oliver Jowett
and Dmitry Tkach
 Modified Files:
  Tag: REL7_3_STABLE
  jdbc/org/postgresql/Driver.java.in
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java

21 years agoNew WAL fix release text is:
Bruce Momjian [Wed, 23 Jul 2003 22:59:39 +0000 (22:59 +0000)]
New WAL fix release text is:

    Prevent rare possibility of server startup failure (Tom)

21 years agoUpdate all FAQ's for 7.3.4.
Bruce Momjian [Wed, 23 Jul 2003 04:13:13 +0000 (04:13 +0000)]
Update all FAQ's for 7.3.4.

21 years agoStamp 7.3.4 release.
Bruce Momjian [Wed, 23 Jul 2003 04:10:16 +0000 (04:10 +0000)]
Stamp 7.3.4 release.

21 years agowups, took out one memset too many ...
Tom Lane [Tue, 22 Jul 2003 19:13:25 +0000 (19:13 +0000)]
wups, took out one memset too many ...

21 years agoBack-patch fix for bugs in pgstat_initstats.
Tom Lane [Tue, 22 Jul 2003 19:00:36 +0000 (19:00 +0000)]
Back-patch fix for bugs in pgstat_initstats.

21 years agoAllow sub[path|ltre] return void result
Teodor Sigaev [Tue, 22 Jul 2003 14:08:58 +0000 (14:08 +0000)]
Allow sub[path|ltre] return void result

21 years agoMark as deprecated
Teodor Sigaev [Tue, 22 Jul 2003 14:04:50 +0000 (14:04 +0000)]
Mark as deprecated

21 years agoFix to prevent SQL injection attacks when calling setObject(int,Object,int)
Barry Lind [Tue, 22 Jul 2003 05:13:05 +0000 (05:13 +0000)]
Fix to prevent SQL injection attacks when calling setObject(int,Object,int)
where the Object is a String and the type is numeric (i.e. INTEGER,LONG,etc).
The fix applies the standard escaping for these values.

 Modified Files:
  Tag: REL7_3_STABLE
  jdbc/org/postgresql/Driver.java.in
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java

21 years agoOh, for crying in a bucket ... relax Assert so that glibc's strxfrm
Tom Lane [Thu, 17 Jul 2003 22:20:22 +0000 (22:20 +0000)]
Oh, for crying in a bucket ... relax Assert so that glibc's strxfrm
does not dump core.

21 years agoWork around buggy strxfrm() present in some Solaris releases.
Tom Lane [Thu, 17 Jul 2003 20:52:42 +0000 (20:52 +0000)]
Work around buggy strxfrm() present in some Solaris releases.

21 years agoFor COMMENT ON DATABASE where database name is unknown or not the current
Tom Lane [Thu, 17 Jul 2003 20:14:09 +0000 (20:14 +0000)]
For COMMENT ON DATABASE where database name is unknown or not the current
database, emit a WARNING and do nothing, rather than raising ERROR.
Per recent discussion in which we concluded this is the best way to deal
with database dumps that are reloaded into a database of a new name.

21 years agoRepair boundary-case bug introduced by patch of two months ago that
Tom Lane [Thu, 17 Jul 2003 16:45:25 +0000 (16:45 +0000)]
Repair boundary-case bug introduced by patch of two months ago that
fixed incorrect initial setting of StartUpID.  The logic in XLogWrite()
expects that Write->curridx is advanced to the next page as soon as
LogwrtResult points to the end of the current page, but StartupXLOG()
failed to make that happen when the old WAL ended exactly on a page
boundary.  Per trouble report from Hannu Krosing.

21 years agoFix a *second* buffer overrun bug in to_ascii(). Grumble.
Tom Lane [Mon, 14 Jul 2003 16:41:46 +0000 (16:41 +0000)]
Fix a *second* buffer overrun bug in to_ascii().  Grumble.

21 years agoDon't generate 'zero' typeids in the output from gen_cross_product.
Tom Lane [Wed, 25 Jun 2003 20:07:54 +0000 (20:07 +0000)]
Don't generate 'zero' typeids in the output from gen_cross_product.
This is no longer necessary or appropriate since we don't use zero typeid
as a wildcard anymore, and it fixes a nasty performance problem with
functions with many parameters.  Per recent example from Reuven Lerner.

21 years ago[ Backpatch to 7.3.X.]
Bruce Momjian [Wed, 25 Jun 2003 04:09:15 +0000 (04:09 +0000)]
[ Backpatch to 7.3.X.]

If they're not, the below causes problems, as the foreign key is added
after the CHECK.  Cluster depends on the index name, so I thought it
wise to ensure all names are available, rather than leaving off the
CONSTRAINT "$n" portion for internally named constraints.

CREATE TABLE jkey (col integer primary key);
CREATE TABLE j (col integer REFERENCES jkey);
ALTER TABLE j ADD CHECK(col > 5);

This is a problem in 7.3 series as well as -Tip.

--
Rod Taylor <rbt@rbt.ca>

21 years agoFix bugs in interval-to-time conversion: HAVE_INT64_TIMESTAMP case did not
Tom Lane [Mon, 16 Jun 2003 18:56:53 +0000 (18:56 +0000)]
Fix bugs in interval-to-time conversion: HAVE_INT64_TIMESTAMP case did not
work at all, and neither case behaved sanely for negative intervals.

21 years agoFix SQL function executor for case where last command of a function is
Tom Lane [Thu, 12 Jun 2003 17:29:37 +0000 (17:29 +0000)]
Fix SQL function executor for case where last command of a function is
not a SELECT.  We didn't use to allow that, but we do now.

21 years agoFix brain damage in deciding which python input converter to use.
Tom Lane [Wed, 11 Jun 2003 18:33:46 +0000 (18:33 +0000)]
Fix brain damage in deciding which python input converter to use.

21 years agoAdd defense in assign_session_authorization() against trying to do
Tom Lane [Fri, 6 Jun 2003 16:25:52 +0000 (16:25 +0000)]
Add defense in assign_session_authorization() against trying to do
catalog lookups when not in a transaction.  This prevents bizarre
failures if someone tries to set a value for session_authorization in
postgresql.conf.  Per report from Fernando Nasser.

21 years agoFix misstatement in release-notes item.
Tom Lane [Mon, 26 May 2003 18:58:36 +0000 (18:58 +0000)]
Fix misstatement in release-notes item.

21 years agoUpdate for 7.3.3. REL7_3_3
Tom Lane [Thu, 22 May 2003 20:38:56 +0000 (20:38 +0000)]
Update for 7.3.3.

21 years agoBrand 7.3.3.
Tom Lane [Thu, 22 May 2003 20:17:24 +0000 (20:17 +0000)]
Brand 7.3.3.

21 years agoUpdate release history for impending 7.3.3 release.
Tom Lane [Thu, 22 May 2003 18:32:01 +0000 (18:32 +0000)]
Update release history for impending 7.3.3 release.

21 years agoRepair sometimes-incorrect computation of StartUpID after a crash, per
Tom Lane [Thu, 22 May 2003 14:39:49 +0000 (14:39 +0000)]
Repair sometimes-incorrect computation of StartUpID after a crash, per
example from Rao Kumar.  This is a very corner corner-case, requiring
a minimum of three closely-spaced database crashes and an unlucky
positioning of the second recovery's checkpoint record before you'd notice
any problem.  But the consequences are dire enough that it's a must-fix.

21 years agoBack-patch Jan's fix to avoid primary key lookup (and lock) if foreign key
Tom Lane [Wed, 21 May 2003 18:14:46 +0000 (18:14 +0000)]
Back-patch Jan's fix to avoid primary key lookup (and lock) if foreign key
does not change on UPDATE.

21 years agoUse -fPIC on Sparc, per Tom Callaway.
Tom Lane [Mon, 19 May 2003 17:51:08 +0000 (17:51 +0000)]
Use -fPIC on Sparc, per Tom Callaway.

21 years agoBack-patch change to avoid O(N^2) behavior with lots of deferred triggers,
Tom Lane [Mon, 19 May 2003 17:23:54 +0000 (17:23 +0000)]
Back-patch change to avoid O(N^2) behavior with lots of deferred triggers,
by making deferredTriggerInvokeEvents only scan events added since it last ran.

21 years agoUpped JDBC build number to 110 for the 7.3.3 release
Barry Lind [Mon, 19 May 2003 04:12:52 +0000 (04:12 +0000)]
Upped JDBC build number to 110 for the 7.3.3 release

 Modified Files:
  Tag: REL7_3_STABLE
  Driver.java.in

21 years agoRepair quoting sloppiness, lack of schema awareness in reindexdb.
Tom Lane [Mon, 19 May 2003 01:00:18 +0000 (01:00 +0000)]
Repair quoting sloppiness, lack of schema awareness in reindexdb.

21 years agoSmall translation updates for 7.3.3 release.
Peter Eisentraut [Sun, 18 May 2003 23:51:41 +0000 (23:51 +0000)]
Small translation updates for 7.3.3 release.

21 years agoFix failure when uniq-ifying an array of zero elements.
Tom Lane [Fri, 16 May 2003 18:50:00 +0000 (18:50 +0000)]
Fix failure when uniq-ifying an array of zero elements.

Teodor Sigaev

21 years agoBack-patch fix to allow createuser to exit on control-C (Oliver Elphick)
Tom Lane [Fri, 16 May 2003 15:17:02 +0000 (15:17 +0000)]
Back-patch fix to allow createuser to exit on control-C (Oliver Elphick)

21 years agoBack-patch fix from Oliver Elphick to force ISO datestyle in dumps.
Tom Lane [Fri, 16 May 2003 13:57:03 +0000 (13:57 +0000)]
Back-patch fix from Oliver Elphick to force ISO datestyle in dumps.

21 years agoSmall changes to use the absolute path to system catalogs.
Tom Lane [Fri, 16 May 2003 13:37:11 +0000 (13:37 +0000)]
Small changes to use the absolute path to system catalogs.

Greg Sabino Mullane

21 years agoCheck calling context for connectby_text(), per Joe Conway.
Tom Lane [Fri, 16 May 2003 06:07:58 +0000 (06:07 +0000)]
Check calling context for connectby_text(), per Joe Conway.

21 years agoApply fixes for problems with dropped columns whose types have also been
Tom Lane [Mon, 12 May 2003 00:17:34 +0000 (00:17 +0000)]
Apply fixes for problems with dropped columns whose types have also been
dropped.  Add regression test, too.

21 years agoAdjust CreateCheckpoint so that buffer dumping activities and cleanup of
Tom Lane [Sat, 10 May 2003 18:01:49 +0000 (18:01 +0000)]
Adjust CreateCheckpoint so that buffer dumping activities and cleanup of
dead xlog segments are not considered part of a critical section.  It is
not necessary to force a database-wide panic if we get a failure in these
operations.  Per recent trouble reports.

21 years agoBe more precise about check for flex 2.5.3, backpatched to 7.3.X.
Bruce Momjian [Wed, 7 May 2003 04:49:02 +0000 (04:49 +0000)]
Be more precise about check for flex 2.5.3, backpatched to 7.3.X.

21 years agoAllow 60 in seconds fields of timestamp, time, interval input values.
Tom Lane [Sun, 4 May 2003 04:30:35 +0000 (04:30 +0000)]
Allow 60 in seconds fields of timestamp, time, interval input values.
Per recent discussion on pgsql-general, this is appropriate for spec
compliance, and has the nice side-effect of easing porting from old
pg_dump files that exhibit the 59.999=>60.000 roundoff problem.

21 years agoWhen a TIMESTAMP, TIME, or INTERVAL precision is specified larger than our
Tom Lane [Sun, 4 May 2003 00:04:10 +0000 (00:04 +0000)]
When a TIMESTAMP, TIME, or INTERVAL precision is specified larger than our
implementation limits, do not issue an ERROR; instead issue a NOTICE and use
the max supported value.  Per pgsql-general discussion of 28-Apr, this is
needed to allow easy porting from pre-7.3 releases where the limits were
higher.

Unrelated change in same area: accept GLOBAL TEMP/TEMPORARY as a synonym
for TEMPORARY, as per pgsql-hackers discussion of 15-Apr.  We previously
rejected it, but that was based on a misreading of the spec --- SQL92's
GLOBAL temp tables are really closer to what we have than their LOCAL ones.

21 years agoFix erroneous space calculation leading to core dump in dumpProcLangs,
Tom Lane [Sat, 3 May 2003 22:19:18 +0000 (22:19 +0000)]
Fix erroneous space calculation leading to core dump in dumpProcLangs,
per report from Olivier Prenant.  Also fix off-by-one space calculation
in ReadToc; this woould not have hurt us until we had more than 100
dependencies for a single object, but wrong is wrong.

21 years agoRepair permissions problem in RI triggers: query parsing has to be done
Tom Lane [Sat, 26 Apr 2003 22:21:58 +0000 (22:21 +0000)]
Repair permissions problem in RI triggers: query parsing has to be done
as the correct user, not only query execution.  Per report from Sean
Chittenden.

21 years agoCorrect oversight in createlang: test for pre-existing handler function
Tom Lane [Sat, 26 Apr 2003 15:19:05 +0000 (15:19 +0000)]
Correct oversight in createlang: test for pre-existing handler function
was broken by opaque->language_handler change.  I see this is already
fixed in CVS tip, but must back-patch for 7.3.3.

21 years agoBack-patch fixes for zero-column tables in COPY, pg_dump.
Tom Lane [Fri, 25 Apr 2003 22:14:33 +0000 (22:14 +0000)]
Back-patch fixes for zero-column tables in COPY, pg_dump.

21 years agoMake [VACUUM] ANALYZE safe on zero-column tables.
Tom Lane [Fri, 25 Apr 2003 21:29:22 +0000 (21:29 +0000)]
Make [VACUUM] ANALYZE safe on zero-column tables.

22 years agoFix misbehavior of func_error() on type names containing '%'.
Tom Lane [Wed, 23 Apr 2003 18:20:10 +0000 (18:20 +0000)]
Fix misbehavior of func_error() on type names containing '%'.
Will patch separately but equivalently in HEAD.

22 years agoFix misbehavior of replace() on strings containing '%'.
Tom Lane [Wed, 23 Apr 2003 18:19:23 +0000 (18:19 +0000)]
Fix misbehavior of replace() on strings containing '%'.
Will patch separately but equivalently in HEAD.

22 years agoPrevent palloc(0) error when parent table has zero columns.
Tom Lane [Mon, 21 Apr 2003 15:20:02 +0000 (15:20 +0000)]
Prevent palloc(0) error when parent table has zero columns.

22 years agoFix char-vs-pg_wchar confusion in p_ere(), per failure report from
Tom Lane [Sun, 20 Apr 2003 17:37:22 +0000 (17:37 +0000)]
Fix char-vs-pg_wchar confusion in p_ere(), per failure report from
Tom O'Dowd.  This fix is not relevant to CVS tip anymore, but we should
fix it in 7.3.*.

22 years agoFix abstime-to-time cast function, which has had broken implementation
Tom Lane [Thu, 17 Apr 2003 04:50:35 +0000 (04:50 +0000)]
Fix abstime-to-time cast function, which has had broken implementation
since 7.2, per bug #947.  Turns out it had wrong volatility label, too.
Can't force initdb in 7.3 branch, but fix anyway for future installs.

22 years agoApplied patches from Kris Jurka fixing a string tokenizing problem and
Barry Lind [Thu, 17 Apr 2003 04:19:55 +0000 (04:19 +0000)]
Applied patches from Kris Jurka fixing a string tokenizing problem and
fixing an order by problem for index metadata results.
Also includes removing some unused code as well as a fix to the toString
method on statement.

 Modified Files:
  Tag: REL7_3_STABLE
  jdbc/org/postgresql/jdbc1/AbstractJdbc1DatabaseMetaData.java
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java

22 years agoFix stupid oversight ...
Tom Lane [Wed, 16 Apr 2003 04:38:05 +0000 (04:38 +0000)]
Fix stupid oversight ...

22 years agoeqjoinsel's logic for case where MCV lists are not present should
Tom Lane [Tue, 15 Apr 2003 05:18:30 +0000 (05:18 +0000)]
eqjoinsel's logic for case where MCV lists are not present should
account for NULLs; in hindsight this is obvious since the code for
the MCV-lists case would reduce to this when there are zero entries
in both lists.  Per example from Alec Mitchell.

22 years agoSecond try at avoiding conflicts with system isblank().
Tom Lane [Sun, 13 Apr 2003 04:07:31 +0000 (04:07 +0000)]
Second try at avoiding conflicts with system isblank().

22 years agoGuard against macro versions of isblank().
Tom Lane [Sat, 12 Apr 2003 22:28:40 +0000 (22:28 +0000)]
Guard against macro versions of isblank().

22 years agoFix encoding conversion function bug.
Tatsuo Ishii [Sat, 12 Apr 2003 08:01:23 +0000 (08:01 +0000)]
Fix encoding conversion function bug.
See following posting for more details.

Subject: Re: [HACKERS] [BUGS] Bug #943: Server-Encoding from EUC_TW to UTF-8 doesn't
From: Tatsuo Ishii <t-ishii@sra.co.jp>
To: michael.enke@wincor-nixdorf.com, pgsql-bugs@postgresql.org
Cc: pgsql-hackers@postgresql.org
Date: Sat, 12 Apr 2003 10:51:45 +0900 (JST)

22 years agoFix error recovery for SSL_read/SSL_write calls.
Tom Lane [Thu, 10 Apr 2003 23:03:13 +0000 (23:03 +0000)]
Fix error recovery for SSL_read/SSL_write calls.

22 years agoRemove premature attempt to constant-fold type coercion expressions.
Tom Lane [Thu, 10 Apr 2003 02:47:57 +0000 (02:47 +0000)]
Remove premature attempt to constant-fold type coercion expressions.
While usually safe, this fails if the coercion function needs the
query snapshot to be set already.  Per example from Nigel Andrews.

22 years agoadded addDataType to the interface
Dave Cramer [Tue, 8 Apr 2003 17:26:39 +0000 (17:26 +0000)]
added addDataType to the interface

22 years agoBack-patch changes to validate page header fields immediately after
Tom Lane [Fri, 4 Apr 2003 00:32:57 +0000 (00:32 +0000)]
Back-patch changes to validate page header fields immediately after
reading in any page.  Also back-port the zero_damaged_pages boolean
that determines what to do about it.