]> granicus.if.org Git - postgresql/log
postgresql
23 years agoHack for Jade vs TeX interpretation of "--"
Peter Eisentraut [Thu, 4 Oct 2001 22:25:48 +0000 (22:25 +0000)]
Hack for Jade vs TeX interpretation of "--"

23 years agoConsider interpreting a function call as a trivial (binary-compatible)
Tom Lane [Thu, 4 Oct 2001 22:06:46 +0000 (22:06 +0000)]
Consider interpreting a function call as a trivial (binary-compatible)
type coercion after failing to find an exact match in pg_proc, but before
considering interpretations that involve a function call with one or
more argument type coercions.  This avoids surprises wherein what looks
like a type coercion is interpreted as coercing to some third type and
then to the destination type, as in Dave Blasby's bug report of 3-Oct-01.
See subsequent discussion in pghackers.

23 years agoOoops, I was a little too enthusiastic about suppressing default
Tom Lane [Thu, 4 Oct 2001 22:00:10 +0000 (22:00 +0000)]
Ooops, I was a little too enthusiastic about suppressing default
index opclasses; they might be default for some other datatype,
in which case we mustn't suppress 'em.

23 years agoPlug memory leaks introduced by dynamic-search-path changes.
Tom Lane [Thu, 4 Oct 2001 19:13:55 +0000 (19:13 +0000)]
Plug memory leaks introduced by dynamic-search-path changes.
From Teodor Sigaev.

23 years agoMake the world safe for atttypmod=0 ... this didn't use to mean anything,
Tom Lane [Thu, 4 Oct 2001 17:52:24 +0000 (17:52 +0000)]
Make the world safe for atttypmod=0 ... this didn't use to mean anything,
but timestamp now wants it to mean something.

23 years agoMake current_time deliver a fractional-second answer.
Tom Lane [Thu, 4 Oct 2001 17:10:11 +0000 (17:10 +0000)]
Make current_time deliver a fractional-second answer.

23 years agoelog command need to use unicode conversion too.
Bruce Momjian [Thu, 4 Oct 2001 15:48:37 +0000 (15:48 +0000)]
elog command need to use unicode conversion too.

Vsevolod Lobko

23 years ago This patch adds reporting of tcl global variable errorInfo
Bruce Momjian [Thu, 4 Oct 2001 15:47:41 +0000 (15:47 +0000)]
  This patch adds reporting of tcl global variable errorInfo
which contains stack trace.
  One problem, after this patch errors will generate multiline ERROR
messages. Is it acceptable or do I need split it and generate multiple
singleline messages?

Vsevolod Lobko

23 years agoAttached is a patch which deals with
Bruce Momjian [Thu, 4 Oct 2001 15:46:49 +0000 (15:46 +0000)]
Attached is a patch which deals with

select 'id' as xxx from table

The issue is:

When the driver gets a data type which does not map into the SQL.Types
it attempts to load the object into a java object. Eventually throwing
an exception indicating that the type "unknown" was not found.

Since the backend defaults "unknown" types to text it was suggested that
the jdbc driver do the same.

This patch does just that.

I have tested it on the above select statement as well as a small
program that serializes, and deserializes a class

Dave Cramer

23 years agoUsing strtol() on int8 values (input parameters or result sets) in
Bruce Momjian [Thu, 4 Oct 2001 15:45:49 +0000 (15:45 +0000)]
Using strtol() on int8 values (input parameters or result sets) in
plpython would result in numeric overflows causing the backend to
terminate abruptly.

This patch fixes it.

Bradley McLean

23 years ago I've just finished the Chinese NLS support(zh_CN part) files
Bruce Momjian [Thu, 4 Oct 2001 15:44:14 +0000 (15:44 +0000)]
I've just finished the Chinese NLS support(zh_CN part) files
translation,
    the attachement is those four files.
   (.tar.gz file, with the directory, root is $PGSRC).
    Tested on some Linux platform.

Weiping He

23 years ago 1. Implemented binary search in array
Bruce Momjian [Thu, 4 Oct 2001 15:41:14 +0000 (15:41 +0000)]
1. Implemented binary search in array

Oleg Bartunov

23 years agoAttached please find an initiation of the
Bruce Momjian [Thu, 4 Oct 2001 15:38:28 +0000 (15:38 +0000)]
Attached please find an initiation of the
NLS effort for the 'postgresql' component.
It includes 'ru.po', which should be added
to /src/backend/po/ and a typical patch for
nls.mk with the 'ru' entry added in
the AVAIL_LANGUAGES.

I'll submit a patch with more translations
for pg_dump once CVSweb becomes available.
CVSweb for me right now is the only way
to get to the source code until my system
is completely setup by the end of the week.

--
Serguei A. Mokhov

23 years agoRemove ChangeLog. No one has defended its existence, and a few people
Bruce Momjian [Thu, 4 Oct 2001 15:28:33 +0000 (15:28 +0000)]
Remove ChangeLog.  No one has defended its existence, and a few people
want it removed.

23 years agoAdjustTimeForTypmod has the same bug ...
Tom Lane [Thu, 4 Oct 2001 15:14:22 +0000 (15:14 +0000)]
AdjustTimeForTypmod has the same bug ...

23 years agoMake regress tests safe for variable-resolution timestamps.
Tom Lane [Thu, 4 Oct 2001 14:51:06 +0000 (14:51 +0000)]
Make regress tests safe for variable-resolution timestamps.

23 years agoFix some problems in new variable-resolution-timestamp code.
Tom Lane [Thu, 4 Oct 2001 14:49:57 +0000 (14:49 +0000)]
Fix some problems in new variable-resolution-timestamp code.

23 years agoChange some 7.2 mentions to version macro.
Bruce Momjian [Thu, 4 Oct 2001 04:46:43 +0000 (04:46 +0000)]
Change some 7.2 mentions to version macro.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 4 Oct 2001 04:24:24 +0000 (04:24 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 4 Oct 2001 04:23:13 +0000 (04:23 +0000)]
Update TODO list.

23 years agoAdd dependency for regexec.c
Tatsuo Ishii [Thu, 4 Oct 2001 04:16:16 +0000 (04:16 +0000)]
Add dependency for regexec.c

23 years agoAdd dependency for like.c
Tatsuo Ishii [Thu, 4 Oct 2001 04:13:40 +0000 (04:13 +0000)]
Add dependency for like.c

23 years agoOptimization for single byte encodings.
Tatsuo Ishii [Thu, 4 Oct 2001 02:15:47 +0000 (02:15 +0000)]
Optimization for single byte encodings.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 22:27:58 +0000 (22:27 +0000)]
Update TODO list.

23 years agoConsistently use gcc's __attribute__((format)) to check sprintf-style
Tom Lane [Wed, 3 Oct 2001 21:58:28 +0000 (21:58 +0000)]
Consistently use gcc's __attribute__((format)) to check sprintf-style
format strings wherever possible.  Remarkably, this exercise didn't
turn up any inconsistencies, but it seems a good idea for the future.

23 years agoDROP AGGREGATE and COMMENT ON AGGREGATE now accept the expected syntax
Tom Lane [Wed, 3 Oct 2001 20:54:22 +0000 (20:54 +0000)]
DROP AGGREGATE and COMMENT ON AGGREGATE now accept the expected syntax
'aggname (aggtype)'.  The old syntax 'aggname aggtype' is still accepted
for backwards compatibility.  Fix pg_dump, which was actually broken for
most cases of user-defined aggregates.  Clean up error messages associated
with these commands.

23 years agoUpdate error message wording for format_type change.
Tom Lane [Wed, 3 Oct 2001 19:24:36 +0000 (19:24 +0000)]
Update error message wording for format_type change.

23 years agoRemove no-longer-used macros.
Tom Lane [Wed, 3 Oct 2001 19:18:42 +0000 (19:18 +0000)]
Remove no-longer-used macros.

23 years agoUpdate incorrect expected file. Use 'timestamp without time zone'
Tom Lane [Wed, 3 Oct 2001 18:39:12 +0000 (18:39 +0000)]
Update incorrect expected file.  Use 'timestamp without time zone'
datatype in test, to try to avoid any dependency on local time zone.

23 years agoUpdate for GIST.
Bruce Momjian [Wed, 3 Oct 2001 18:38:06 +0000 (18:38 +0000)]
Update for GIST.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 18:36:03 +0000 (18:36 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 18:35:41 +0000 (18:35 +0000)]
Update TODO list.

23 years agoTeach format_type about time/timestamp precision, so that pg_dump and
Tom Lane [Wed, 3 Oct 2001 18:32:42 +0000 (18:32 +0000)]
Teach format_type about time/timestamp precision, so that pg_dump and
psql's \d option work properly.

23 years agoAdd missing TIMESTAMPTZ case.
Tom Lane [Wed, 3 Oct 2001 18:25:59 +0000 (18:25 +0000)]
Add missing TIMESTAMPTZ case.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 17:38:41 +0000 (17:38 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 17:38:13 +0000 (17:38 +0000)]
Update TODO list.

23 years agoSome further kibitzing on cachability of datetime functions.
Tom Lane [Wed, 3 Oct 2001 17:22:05 +0000 (17:22 +0000)]
Some further kibitzing on cachability of datetime functions.
In particular, mark all I/O functions that can produce or consume
a date as noncachable, in light of existence of DateStyle variable.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 3 Oct 2001 16:05:43 +0000 (16:05 +0000)]
Update TODO list.

23 years agoUpdate for latest regression tests.
Tom Lane [Wed, 3 Oct 2001 15:51:48 +0000 (15:51 +0000)]
Update for latest regression tests.

23 years agoAdjustTimestampForTypmod does not work (at least not portably) on
Tom Lane [Wed, 3 Oct 2001 15:50:48 +0000 (15:50 +0000)]
AdjustTimestampForTypmod does not work (at least not portably) on
-infinity and +infinity.  Put TIMESTAMP_NOT_FINITE guard into the routine,
instead of forgetting it at some call sites.  Fixes regression test
failures here.

23 years agoNeatnik fetishism.
Tom Lane [Wed, 3 Oct 2001 15:42:12 +0000 (15:42 +0000)]
Neatnik fetishism.

23 years agoImplement precision support for timestamp and time, both with and without
Thomas G. Lockhart [Wed, 3 Oct 2001 05:29:27 +0000 (05:29 +0000)]
Implement precision support for timestamp and time, both with and without
 time zones.
SQL99 spec requires a default of zero (round to seconds) which is set
 in gram.y as typmod is set in the parse tree. We *could* change to a
 default of either 6 (for internal compatibility with previous versions)
 or 2 (for external compatibility with previous versions).
Evaluate entries in pg_proc wrt the iscachable attribute for timestamp and
 other date/time types. Try to recognize cases where side effects like the
 current time zone setting may have an effect on results to decide whether
 something is cachable or not.

23 years agoMinor fixup in spacing for tabular information.
Thomas G. Lockhart [Wed, 3 Oct 2001 05:24:22 +0000 (05:24 +0000)]
Minor fixup in spacing for tabular information.

23 years agoFix up bad indenting in a few places.
Thomas G. Lockhart [Wed, 3 Oct 2001 05:23:12 +0000 (05:23 +0000)]
Fix up bad indenting in a few places.

23 years agoTweak to test HAVE_GETOPT_H before including <getopt.h>, so as to
Tom Lane [Tue, 2 Oct 2001 22:38:43 +0000 (22:38 +0000)]
Tweak to test HAVE_GETOPT_H before including <getopt.h>, so as to
build on systems without.

23 years agoAdd CREATE OR REPLACE FUNCTION syntax to allow replacing a function
Tom Lane [Tue, 2 Oct 2001 21:39:36 +0000 (21:39 +0000)]
Add CREATE OR REPLACE FUNCTION syntax to allow replacing a function
definition without changing the function's OID, thereby not breaking
rules, views, triggers, etc that depend on it.  From Gavin Sherry.

23 years agoUpdated URLs.
Vince Vielhaber [Tue, 2 Oct 2001 14:21:26 +0000 (14:21 +0000)]
Updated URLs.

23 years agoRe-added Tom's patch fixing my setlocale patch. I accidently
Michael Meskes [Tue, 2 Oct 2001 14:08:28 +0000 (14:08 +0000)]
Re-added Tom's patch fixing my setlocale patch. I accidently
deleted it.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 2 Oct 2001 01:21:16 +0000 (01:21 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 2 Oct 2001 01:20:46 +0000 (01:20 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 2 Oct 2001 01:20:14 +0000 (01:20 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 2 Oct 2001 01:20:05 +0000 (01:20 +0000)]
Update TODO list.

23 years agoRevise shmget() and semget() failure messages to mention the possibility
Tom Lane [Mon, 1 Oct 2001 23:26:55 +0000 (23:26 +0000)]
Revise shmget() and semget() failure messages to mention the possibility
of coping by reducing shared_buffers/max_connections settings.

23 years agoFix typo (block size is 8 kB not 8192 kB).
Tom Lane [Mon, 1 Oct 2001 22:44:31 +0000 (22:44 +0000)]
Fix typo (block size is 8 kB not 8192 kB).

23 years agoFix wording.
Bruce Momjian [Mon, 1 Oct 2001 21:47:24 +0000 (21:47 +0000)]
Fix wording.

23 years agoChange pg_dump to produce CREATE INDEX commands by using the backend's
Tom Lane [Mon, 1 Oct 2001 21:31:52 +0000 (21:31 +0000)]
Change pg_dump to produce CREATE INDEX commands by using the backend's
pg_get_indexdef() function, rather than reaching into the system catalogs
for itself.  This eliminates a fair amount of redundant code.  Also,
since I just changed pg_get_indexdef() to suppress display of default
index opclasses, this will mean that 7.2 and later dumps will not mention
opclasses unless they are non-default opclasses.  Should make life easier
for future index opclass reorganizations.

23 years agoExtend pg_get_indexdef() to know about index predicates. Also, tweak
Tom Lane [Mon, 1 Oct 2001 20:15:26 +0000 (20:15 +0000)]
Extend pg_get_indexdef() to know about index predicates.  Also, tweak
it to suppress index opclass output for opclasses that are the default
for their datatype; only non-default opclasses are shown explicitly.
This is expected to improve portability of the CREATE INDEX command
across future versions of Postgres --- we've changed index opclasses
too often in the past to think we won't do so again.

23 years agoMake inclusion logic for sys/sem.h and sys/ipc.h consistent across all
Tom Lane [Mon, 1 Oct 2001 18:16:35 +0000 (18:16 +0000)]
Make inclusion logic for sys/sem.h and sys/ipc.h consistent across all
the files that need them.  Per trouble report from Teodor.

23 years agoFix a couple of stray // comments.
Tom Lane [Mon, 1 Oct 2001 17:53:12 +0000 (17:53 +0000)]
Fix a couple of stray // comments.

23 years agoAdd sys/types.h for FreeBSD compile.
Bruce Momjian [Mon, 1 Oct 2001 17:52:34 +0000 (17:52 +0000)]
Add sys/types.h for FreeBSD compile.

Teodor Sigaev

23 years agoUpdate docs for 7.2 mention where appropriate.
Bruce Momjian [Mon, 1 Oct 2001 17:46:46 +0000 (17:46 +0000)]
Update docs for 7.2 mention where appropriate.

23 years agoClean up a little more carefully around fork of pgstat subprocess.
Tom Lane [Mon, 1 Oct 2001 16:48:37 +0000 (16:48 +0000)]
Clean up a little more carefully around fork of pgstat subprocess.
Partial support for BEOS (not sure whether second fork of grandchild
process needs these extra calls or not; someone who has BEOS will need
to test it).

23 years agoAdd missing pgcrypto files.
Bruce Momjian [Mon, 1 Oct 2001 16:12:23 +0000 (16:12 +0000)]
Add missing pgcrypto files.

23 years agoUpdate to "7.2".
Bruce Momjian [Mon, 1 Oct 2001 15:44:45 +0000 (15:44 +0000)]
Update to "7.2".

23 years agoRun autoconf.
Bruce Momjian [Mon, 1 Oct 2001 15:33:55 +0000 (15:33 +0000)]
Run autoconf.

23 years ago> > > > > - PostgreSQL requires to be compiled with --enable-multibyte
Bruce Momjian [Mon, 1 Oct 2001 15:33:31 +0000 (15:33 +0000)]
> > > > > - PostgreSQL requires to be compiled with --enable-multibyte
> > > > >   and --enable-unicode-convertion if it ought to work correctly
> > > > >   with Tcl/Tk >= 8.1 (client or server side).
> > > > >
> > > > > - PL/Tcl needs to be changed to use pg_do_encoding_conversion
> > > > >   if it runs on a Tcl version >= 8.1 .
> > >
> > > > I'll do pl/tcl part in the next version of patch. Using this approach we
> > > > can eliminate overhead for databases in UNICODE.
> > >
> > > Any progress on this?  I'd prefer to get rid of this --enable-pltcl-utf
> > > option before release.
> >
> > Done
> >
> > Next version removes --enable-pltcl-utf switch and enables embedded
> > utf conversion of pgsql if tcl version >=8.1 and --enable-unicode-conversion

23 years ago - Fixed truncate bug.
Michael Meskes [Mon, 1 Oct 2001 12:02:28 +0000 (12:02 +0000)]
- Fixed truncate bug.
        - Added patch by Christof Petig <christof.petig@wtal.de> to
clean up
ecpglib.

23 years agoUpdate multibyte Japanese doc
Tatsuo Ishii [Mon, 1 Oct 2001 08:16:38 +0000 (08:16 +0000)]
Update multibyte Japanese doc

23 years agoAnother round of cleanups for dynahash.c (maybe it's finally clean of
Tom Lane [Mon, 1 Oct 2001 05:36:17 +0000 (05:36 +0000)]
Another round of cleanups for dynahash.c (maybe it's finally clean of
portability issues).  Caller-visible data structures are now allocated
on MAXALIGN boundaries, allowing safe use of datatypes wider than 'long'.
Rejigger hash_create API so that caller specifies size of key and
total size of entry, not size of key and size of rest of entry.
This simplifies life considerably since each number is just a sizeof(),
and padding issues etc. are taken care of automatically.

23 years agoSuppress timestamp_ops for backwards compatibility with 7.1 pg_dump.
Tom Lane [Mon, 1 Oct 2001 04:19:18 +0000 (04:19 +0000)]
Suppress timestamp_ops for backwards compatibility with 7.1 pg_dump.

23 years agoRemove C++ style comment. Some compilers won't accept it.
Tatsuo Ishii [Mon, 1 Oct 2001 02:31:33 +0000 (02:31 +0000)]
Remove C++ style comment. Some compilers won't accept it.

23 years agoFix a compile error on a platform that does not have PAM.
Tatsuo Ishii [Mon, 1 Oct 2001 02:26:36 +0000 (02:26 +0000)]
Fix a compile error on a platform that does not have PAM.

23 years agoAdd pgstattuple
Tatsuo Ishii [Mon, 1 Oct 2001 01:52:38 +0000 (01:52 +0000)]
Add pgstattuple

23 years agoMake libpq++ safe again for older C++ compilers. Do 'using namespace std'
Tom Lane [Sun, 30 Sep 2001 22:30:37 +0000 (22:30 +0000)]
Make libpq++ safe again for older C++ compilers.  Do 'using namespace std'
only if configure found it was safe to do so; do not assume const_cast
is available.

23 years ago* regression tests
Bruce Momjian [Sun, 30 Sep 2001 22:18:29 +0000 (22:18 +0000)]
* regression tests
* minor doc updates

Marko Kreen

23 years agoThese are further fixes for double quotes missing in the various shell
Bruce Momjian [Sun, 30 Sep 2001 22:17:51 +0000 (22:17 +0000)]
These are further fixes for double quotes missing in the various shell
scripts.

Justin Clift

23 years agoClean up encode/decode functions a little bit.
Tom Lane [Sun, 30 Sep 2001 22:03:41 +0000 (22:03 +0000)]
Clean up encode/decode functions a little bit.

23 years agoFix sloppiness about static vs non-static declaration of functions.
Tom Lane [Sun, 30 Sep 2001 21:48:58 +0000 (21:48 +0000)]
Fix sloppiness about static vs non-static declaration of functions.
Some compilers are pickier about this than gcc is.

23 years agopg_stat_get_backend_idset should reset fmgr_info->fn_extra at end of
Tom Lane [Sun, 30 Sep 2001 21:01:39 +0000 (21:01 +0000)]
pg_stat_get_backend_idset should reset fmgr_info->fn_extra at end of
execution, so that it restarts correctly if query tree is used again.

23 years agoAllow the postmaster to accept changes in PGC_BACKEND GUC variables
Tom Lane [Sun, 30 Sep 2001 20:16:21 +0000 (20:16 +0000)]
Allow the postmaster to accept changes in PGC_BACKEND GUC variables
from the config file, so that these changes will propagate to backends
started later.  Already-started backends continue to ignore changes
in these variables.

23 years agoISTM that IsUnderPostmaster should mean we are a child process of the
Tom Lane [Sun, 30 Sep 2001 20:08:18 +0000 (20:08 +0000)]
ISTM that IsUnderPostmaster should mean we are a child process of the
postmaster ... it should not be set in the postmaster itself.

23 years agoChange tar -T file to tar `cat file` for BSD tar.
Bruce Momjian [Sun, 30 Sep 2001 19:18:29 +0000 (19:18 +0000)]
Change tar -T file to tar `cat file` for BSD tar.

23 years agoChange tar -c -f to -cf for BSD tar.
Bruce Momjian [Sun, 30 Sep 2001 19:09:57 +0000 (19:09 +0000)]
Change tar -c -f to -cf for BSD tar.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 30 Sep 2001 19:01:44 +0000 (19:01 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 30 Sep 2001 19:01:18 +0000 (19:01 +0000)]
Update TODO list.

23 years agoCreate a GUC parameter max_files_per_process that is a configurable
Tom Lane [Sun, 30 Sep 2001 18:57:45 +0000 (18:57 +0000)]
Create a GUC parameter max_files_per_process that is a configurable
upper limit on what we will believe from sysconf(_SC_OPEN_MAX).  The
default value is 1000, so that under ordinary conditions it won't
affect the behavior.  But on platforms where the kernel promises far
more than it can deliver, this can be used to prevent running out of
file descriptors.  See numerous past discussions, eg, pgsql-hackers
around 23-Dec-2000.

23 years agoAddition of xid=int4 operator makes opr_sanity unhappy.
Tom Lane [Sun, 30 Sep 2001 17:37:32 +0000 (17:37 +0000)]
Addition of xid=int4 operator makes opr_sanity unhappy.

23 years agowinsock_strerror crashed on me. This fixes a line of code that looks
Tom Lane [Sun, 30 Sep 2001 16:23:30 +0000 (16:23 +0000)]
winsock_strerror crashed on me. This fixes a line of code that looks
cleverer than it actually is ;-) Braces are good for you :-)

Gerhard HÃŒring

23 years agoChanges:
Tom Lane [Sun, 30 Sep 2001 16:16:44 +0000 (16:16 +0000)]
Changes:

 1. gist__int_ops is now without lossy
 2. added sort entry in picksplit

Oleg Bartunov

23 years agoUpdate required due to recent changes in FigureColname.
Tom Lane [Sun, 30 Sep 2001 16:15:59 +0000 (16:15 +0000)]
Update required due to recent changes in FigureColname.

23 years agoUse <mediaobject> instead of <graphic> for forward compatibility. Be more
Peter Eisentraut [Sun, 30 Sep 2001 16:05:54 +0000 (16:05 +0000)]
Use <mediaobject> instead of <graphic> for forward compatibility.  Be more
flexible about the extension of the graphic files, allow for other formats
in print output.  (Generating these formats is not implemented yet.)

23 years agoFix *.gif expansion to be empty when no gif files are used.
Peter Eisentraut [Sun, 30 Sep 2001 16:01:12 +0000 (16:01 +0000)]
Fix *.gif expansion to be empty when no gif files are used.

23 years agoAdd an operator xid '=' int and remove BINARY_COMPATI...
Hiroshi Inoue [Sun, 30 Sep 2001 06:46:58 +0000 (06:46 +0000)]
Add an operator xid '=' int and remove BINARY_COMPATI...

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 30 Sep 2001 00:50:05 +0000 (00:50 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 30 Sep 2001 00:49:15 +0000 (00:49 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Sun, 30 Sep 2001 00:46:15 +0000 (00:46 +0000)]
Update TODO list.

23 years agoBack out change. Too many place to change too close to beta:
Bruce Momjian [Sun, 30 Sep 2001 00:45:48 +0000 (00:45 +0000)]
Back out change.  Too many place to change too close to beta:

* HOLDER/HOLDERTAB rename to PROCLOCKLINK/PROCLOCKLINKTAG (Bruce)

Will return later.

23 years agoTweak btree page split logic so that when splitting a page that is
Tom Lane [Sat, 29 Sep 2001 23:49:51 +0000 (23:49 +0000)]
Tweak btree page split logic so that when splitting a page that is
rightmost on its tree level, we split 2/3 to the left and 1/3 to the
new right page, rather than the even split we use elsewhere.  The idea
is that when faced with a steadily increasing series of inserted keys
(such as sequence or timestamp values), we'll end up with a btree that's
about 2/3ds full not 1/2 full, which is much closer to the desired
steady-state load for a btree.  Per suggestion from Ann Harrison of
IBPhoenix.

23 years agoAllow comparison between xid and xid, int.
Hiroshi Inoue [Sat, 29 Sep 2001 23:01:26 +0000 (23:01 +0000)]
Allow comparison between xid and xid, int.

23 years agoUpdate TODO list.
Bruce Momjian [Sat, 29 Sep 2001 21:37:06 +0000 (21:37 +0000)]
Update TODO list.