]> granicus.if.org Git - postgresql/log
postgresql
25 years agoFrom: Magnus Hagander <mha@sollentuna.net>
Marc G. Fournier [Tue, 12 Jan 1999 12:49:52 +0000 (12:49 +0000)]
From: Magnus Hagander <mha@sollentuna.net>

Here's another patch for the libpq backend areas. This patch removes all
usage of "FILE *" on the communications channel. It also cleans up the
comments and headers in the pqcomm.c file - a lot of things were either
missing or incorrect. Finally, it removes a couple of unused functions
(leftovers from the time of shared code between the libpq backend and
frontend).

25 years agoFrom: Magnus Hagander <mha@sollentuna.net>
Marc G. Fournier [Mon, 11 Jan 1999 03:56:11 +0000 (03:56 +0000)]
From: Magnus Hagander <mha@sollentuna.net>

Here is a first patch to cleanup the backend side of libpq.
This patch removes all external dependencies on the "Pfin" and "Pfout" that
are declared in pqcomm.h. These variables are also changed to "static" to
make sure.
Almost all the change is in the handler of the "copy" command - most other
areas of the backend already used the correct functions.
This change will make the way for cleanup of the internal stuff there - now
that all the functions accessing the file descriptors are confined to a
single directory.

25 years agoBe more careful to check input string lengths as well as values
Thomas G. Lockhart [Sun, 10 Jan 1999 17:20:54 +0000 (17:20 +0000)]
Be more careful to check input string lengths as well as values
 when deciding whether a field is a year field.  Assume *anything* longer
 than 2 digits (if it isn't a special-case doy) is a valid year.
 This should fix the "Y1K" and "Y10K" problems
  pointed out by Massimo recently.
Check usage of BC to require a positive-valued year; before just used it
 to flip the sign of the year without checking. This led to problems
 near year zero.
Allow a 5 digit "concatenated date" of 2 digit year plus day of year.
Do 2->4 digit year correction for 6 and 5 digit "concatenated dates".
 Somehow forgot this originally. Guess not many folks use it...

25 years agoHandle "NaN" and "Infinity" for input values.
Thomas G. Lockhart [Sun, 10 Jan 1999 17:13:06 +0000 (17:13 +0000)]
Handle "NaN" and "Infinity" for input values.
I think NAN is already guaranteed to be there from Jan's work on NUMERIC,
 but perhaps HUGE_VAL needs some #ifndef's in the same place.
Should also include "-Infinity" as -HUGE_VAL sometime; not there yet.

25 years agoUpdate 06-40-0004 -- Add Bookmark support!
Byron Nikolaidis [Fri, 8 Jan 1999 18:24:45 +0000 (18:24 +0000)]
Update 06-40-0004 -- Add Bookmark support!

25 years agoSorry for posting it here again, but I haven't corrected my
Bruce Momjian [Thu, 7 Jan 1999 19:11:18 +0000 (19:11 +0000)]
Sorry for posting it here again, but I haven't corrected my
subscriptions
yet. It's just a small patch to ecpg to keep it in sync with gram.y.

Michael

25 years agoFix build rules for the tutorial,
Thomas G. Lockhart [Thu, 7 Jan 1999 06:19:17 +0000 (06:19 +0000)]
Fix build rules for the tutorial,
 which was not built last time these things changed.

25 years agoRefresh info for various platforms. Claim client-side support for WIN32.
Thomas G. Lockhart [Thu, 7 Jan 1999 03:17:46 +0000 (03:17 +0000)]
Refresh info for various platforms. Claim client-side support for WIN32.

25 years agoFix typo.
Thomas G. Lockhart [Thu, 7 Jan 1999 03:05:28 +0000 (03:05 +0000)]
Fix typo.

25 years agoAdd entries for introduction chapter components such as Y2K statement.
Thomas G. Lockhart [Thu, 7 Jan 1999 03:05:01 +0000 (03:05 +0000)]
Add entries for introduction chapter components such as Y2K statement.
Use a <citetitle> rather than <xref> when referring to libpq
 since that chapter is not included in the tutorial.

25 years agoRemove mention of a couple of introductory chapters
Thomas G. Lockhart [Thu, 7 Jan 1999 03:02:27 +0000 (03:02 +0000)]
Remove mention of a couple of introductory chapters
 which are not used here.

25 years agoFix column formatting for a table. No content changes.
Thomas G. Lockhart [Thu, 7 Jan 1999 03:01:27 +0000 (03:01 +0000)]
Fix column formatting for a table. No content changes.

25 years agoPut in more markup. Not done yet though.
Thomas G. Lockhart [Thu, 7 Jan 1999 03:00:43 +0000 (03:00 +0000)]
Put in more markup. Not done yet though.

25 years agoInclude installation instructions for WIN32 client-side libraries.
Thomas G. Lockhart [Thu, 7 Jan 1999 02:59:45 +0000 (02:59 +0000)]
Include installation instructions for WIN32 client-side libraries.
From Magnus Hagander <mha@sollentuna.net>.

25 years agoUpdate 06-40-0003
Byron Nikolaidis [Wed, 6 Jan 1999 20:44:08 +0000 (20:44 +0000)]
Update 06-40-0003

25 years agoSELECT FOR UPDATE syntax
Vadim B. Mikheev [Tue, 5 Jan 1999 15:46:25 +0000 (15:46 +0000)]
SELECT FOR UPDATE syntax

25 years agoRestricted maximum precision for NUMERIC to 1000 digits.
Jan Wieck [Tue, 5 Jan 1999 11:12:11 +0000 (11:12 +0000)]
Restricted maximum precision for NUMERIC to 1000 digits.
Anything else is CPU overkill.

Jan

25 years agoRemoved precision restriction in numeric_round() causing
Jan Wieck [Tue, 5 Jan 1999 11:10:45 +0000 (11:10 +0000)]
Removed precision restriction in numeric_round() causing
overflow error on high precision calculations where temporary
huge precision is required.

Jan

25 years agoMini Update #2 -- final fixes for buffer lengths, null buffers, truncation
Byron Nikolaidis [Tue, 5 Jan 1999 00:32:21 +0000 (00:32 +0000)]
Mini Update #2 -- final fixes for buffer lengths, null buffers, truncation

25 years agoJust one more bugfix - this time in cmp_abs (cannot imagine that
Jan Wieck [Mon, 4 Jan 1999 17:51:58 +0000 (17:51 +0000)]
Just one more bugfix - this time in cmp_abs (cannot imagine that
taking a logarithm with a 400 digit precision worked with that bug
in place).

Jan

25 years agoFixed backend crashing bug in apply_typmod()
Jan Wieck [Mon, 4 Jan 1999 12:53:23 +0000 (12:53 +0000)]
Fixed backend crashing bug in apply_typmod()

Jan

25 years agoLittle fix for round() function
Jan Wieck [Mon, 4 Jan 1999 11:20:33 +0000 (11:20 +0000)]
Little fix for round() function

Jan

25 years agoupdate HISTORY
Bruce Momjian [Sun, 3 Jan 1999 06:57:05 +0000 (06:57 +0000)]
update HISTORY

25 years agoFix NAN code.
Bruce Momjian [Sun, 3 Jan 1999 05:30:47 +0000 (05:30 +0000)]
Fix NAN code.

25 years agoFix for NAN generation.
Bruce Momjian [Sun, 3 Jan 1999 02:40:12 +0000 (02:40 +0000)]
Fix for NAN generation.

25 years agoMention 6.4.2 changes.
Bruce Momjian [Sat, 2 Jan 1999 16:06:42 +0000 (16:06 +0000)]
Mention 6.4.2 changes.

25 years ago Some security, since we now have vsnprintf, I remade an old patch
Bruce Momjian [Fri, 1 Jan 1999 04:48:49 +0000 (04:48 +0000)]
   Some security, since we now have vsnprintf, I remade an old patch
   with some extra ugly sprintfs fixed. More work in this area is
   needed still.

Göran Thyni

25 years agoFix for no platform NAN.
Bruce Momjian [Fri, 1 Jan 1999 04:17:13 +0000 (04:17 +0000)]
Fix for no platform NAN.

25 years agoAdd max oid display, rather than returning just a dash for the last entry.
Bruce Momjian [Thu, 31 Dec 1998 20:09:49 +0000 (20:09 +0000)]
Add max oid display, rather than returning just a dash for the last entry.

25 years agoChange ordering of HAVE_TM_ZONE and HAVE_INT_TIMEZONE code blocks
Thomas G. Lockhart [Thu, 31 Dec 1998 16:30:59 +0000 (16:30 +0000)]
Change ordering of HAVE_TM_ZONE and HAVE_INT_TIMEZONE code blocks
 to give HAVE_TM_ZONE priority. This fixes glibc2 machines and any other
 machine which passes both tests in configure.
Repair HAVE_TM_ZONE code which stuffs tm structure with date type values.
 Same problems as were originally there before v6.1, but never noticed.
Thanks to Oleg for nagging :)

25 years agouncomment MYLOG to prevent warnings in compile
Byron Nikolaidis [Thu, 31 Dec 1998 00:26:44 +0000 (00:26 +0000)]
uncomment MYLOG to prevent warnings in compile

25 years agoMini update to fix SQLGetInfo to work properly (truncation, NULL)
Byron Nikolaidis [Thu, 31 Dec 1998 00:26:06 +0000 (00:26 +0000)]
Mini update to fix SQLGetInfo to work properly (truncation, NULL)

25 years agoLittle precision fix for POWER(). I discovered problems with big
Jan Wieck [Wed, 30 Dec 1998 20:46:06 +0000 (20:46 +0000)]
Little precision fix for POWER(). I discovered problems with big
exponents.

Jan

25 years agoAdded NUMERIC data type with many builtin funcitons, operators
Jan Wieck [Wed, 30 Dec 1998 19:56:35 +0000 (19:56 +0000)]
Added NUMERIC data type with many builtin funcitons, operators
and aggregates.

Jan

25 years agoUpdate for deadlock detection.
Bruce Momjian [Tue, 29 Dec 1998 19:32:08 +0000 (19:32 +0000)]
Update for deadlock detection.

25 years agoUpdate for deadlock timer.
Bruce Momjian [Tue, 29 Dec 1998 18:36:29 +0000 (18:36 +0000)]
Update for deadlock timer.

25 years agoFix for deadlock detection timeout.
Bruce Momjian [Tue, 29 Dec 1998 18:30:33 +0000 (18:30 +0000)]
Fix for deadlock detection timeout.

25 years agoFix for deadlock timer timeout.
Bruce Momjian [Tue, 29 Dec 1998 18:29:18 +0000 (18:29 +0000)]
Fix for deadlock timer timeout.

25 years agoMinor fixes to compile on unix for v6-40-0002
Byron Nikolaidis [Tue, 29 Dec 1998 14:59:30 +0000 (14:59 +0000)]
Minor fixes to compile on unix for v6-40-0002

25 years agoClean up to ensure tag completion as required by the newest versions
Thomas G. Lockhart [Tue, 29 Dec 1998 02:24:47 +0000 (02:24 +0000)]
Clean up to ensure tag completion as required by the newest versions
 of Norm's Modular Style Sheets and jade/docbook.
From Vince Vielhaber <vev@michvhf.com>.

25 years agoUpdate for version 6-40-0002 and re-merge Thomas' changes.
Byron Nikolaidis [Tue, 29 Dec 1998 01:49:58 +0000 (01:49 +0000)]
Update for version 6-40-0002 and re-merge Thomas' changes.

25 years agoautoconf
Bruce Momjian [Mon, 28 Dec 1998 05:41:51 +0000 (05:41 +0000)]
autoconf

25 years agoFix for current dir not in path, from stan's report.
Bruce Momjian [Mon, 28 Dec 1998 05:37:50 +0000 (05:37 +0000)]
Fix for current dir not in path, from stan's report.

25 years agoLong awaited port for NetBSD/m68k was finally done by Mr. Mutsuki
Bruce Momjian [Sat, 26 Dec 1998 18:15:53 +0000 (18:15 +0000)]
Long awaited port for NetBSD/m68k was finally done by Mr. Mutsuki
Nakajima. Since he is not subscribing the mailing list, I'm posting
his patches by his request.  According to him, he has successfully
compiled and passed the regression test on Mac SE/30 running
NetBSD/m68k. Also, another person has reported that with the patches
PostgreSQL is working on NetBSD/sun3 too.
--
Tatsuo Ishii

25 years agoFix for version, update to 6.5.
Bruce Momjian [Fri, 25 Dec 1998 02:20:12 +0000 (02:20 +0000)]
Fix for version, update to 6.5.

25 years agoFix for snprintf and long long unsigned.
Bruce Momjian [Thu, 24 Dec 1998 05:28:50 +0000 (05:28 +0000)]
Fix for snprintf and long long unsigned.

25 years agoUpdate HISTORY in CURRENT tree.
Bruce Momjian [Thu, 24 Dec 1998 05:21:59 +0000 (05:21 +0000)]
Update HISTORY in CURRENT tree.

25 years agoFix reference to null pointer when no aggregate function candidates
Thomas G. Lockhart [Wed, 23 Dec 1998 14:38:40 +0000 (14:38 +0000)]
Fix reference to null pointer when no aggregate function candidates
 are available. Problem reported by David Sauer <davids@iol.cz>.
Modify information in resulting error message to indicate both agg name
 and data type.

25 years agoAdd subversion mention
Bruce Momjian [Wed, 23 Dec 1998 14:35:42 +0000 (14:35 +0000)]
Add subversion mention

25 years agoFrom: Michael Meskes <Michael.Meskes@usa.net>
Marc G. Fournier [Tue, 22 Dec 1998 18:50:56 +0000 (18:50 +0000)]
From: Michael Meskes <Michael.Meskes@usa.net>

+
+Wed Dec  9 11:24:54 MEZ 1998
+
+       - Synced preproc.y with gram.y and the keywords.c files to add CASE
+         statement.
+
+Tue Dec 22 14:16:11 CET 1998
+
+       - Synced preproc.y with gram.y for locking statements.
+       - Set version to 2.4.5

25 years agoChanged TypeName.typmod to int32 - atttypmod is of that size
Jan Wieck [Mon, 21 Dec 1998 12:50:29 +0000 (12:50 +0000)]
Changed TypeName.typmod to int32 - atttypmod is of that size

Jan

25 years agofinished cleaning up outnodes.c ...
Marc G. Fournier [Sun, 20 Dec 1998 07:13:36 +0000 (07:13 +0000)]
finished cleaning up outnodes.c ...

25 years agoFix deadlock so it only checks once.
Bruce Momjian [Fri, 18 Dec 1998 19:45:38 +0000 (19:45 +0000)]
Fix deadlock so it only checks once.

25 years agoIncluded are patches for doc/README.mb and README.mb.jp.
Bruce Momjian [Fri, 18 Dec 1998 18:01:10 +0000 (18:01 +0000)]
Included are patches for doc/README.mb and README.mb.jp.
Please apply to both trees.
--
Tatsuo Ishii

25 years agoAdd pgaccess Makefile.
Bruce Momjian [Fri, 18 Dec 1998 17:54:42 +0000 (17:54 +0000)]
Add pgaccess Makefile.

25 years agoHere are descriptions:
Bruce Momjian [Fri, 18 Dec 1998 17:53:22 +0000 (17:53 +0000)]
Here are descriptions:

- the first patch is just to preven listing the perl warning in the
  make output unless it is actually emitted by the make.  this may
  prevent new users from being confused by the warning in their output

- the second patch (to 2 files) just enables building/installing
  pgaccess if TCL and TK are available.  a Makefile is created to do
  this, but you may wish to change the heading information in it since
  I just copied another Makefile to use as a template.

I hope these make it into 6.4.1.

Cheers,
Brook

25 years agoI have ported PostgreSQL 6.4 to Cobalt Qube running Linux 2.0.33 with
Bruce Momjian [Fri, 18 Dec 1998 17:25:41 +0000 (17:25 +0000)]
I have ported PostgreSQL 6.4 to Cobalt Qube running Linux 2.0.33 with
MIPS cpu (I think it's R4000). I have tested the patches on LinuxPPC
and FreeBSD. I believe they do not harm other platforms.
--
Tatsuo Ishii

25 years agoFix typos in rules.sgml.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:23:34 +0000 (16:23 +0000)]
Fix typos in rules.sgml.
Add Emacs configuration block to postgres.sgml.

25 years agoInclude example of Postgres' extended query syntax which allow one to omit
Thomas G. Lockhart [Fri, 18 Dec 1998 16:22:31 +0000 (16:22 +0000)]
Include example of Postgres' extended query syntax which allow one to omit
 the FROM clause. e.g. "SELECT table.*".

25 years agoUpdate information on alternate database locations.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:20:28 +0000 (16:20 +0000)]
Update information on alternate database locations.
These fixes are from Jeff Hoffmann <jeff@remapcorp.com>.

25 years agoInclude more details on editing with Emacs.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:17:29 +0000 (16:17 +0000)]
Include more details on editing with Emacs.
Remove mention of the old "migration" flat files.
Change URLs for resources to point to areas, not particular files.
 That way things stay correct even when version of tools change.
 Suggested by Vince Vielhaber.

25 years agoInclude keywords from CASE implementation.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:13:45 +0000 (16:13 +0000)]
Include keywords from CASE implementation.
Rearrange a few other keywords to put them in the proper category.

25 years agoAdd Cobalt MIPS linux box from Tatsuo.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:12:13 +0000 (16:12 +0000)]
Add Cobalt MIPS linux box from Tatsuo.

25 years agoUpdate some info, especially the CIDR/INET data types.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:11:12 +0000 (16:11 +0000)]
Update some info, especially the CIDR/INET data types.

25 years agoInclude Emacs DocBook DTD description file. May allow Emacs to understand
Thomas G. Lockhart [Fri, 18 Dec 1998 16:10:00 +0000 (16:10 +0000)]
Include Emacs DocBook DTD description file. May allow Emacs to understand
 tags even if DocBook is not installed on your local system.
Most of the .sgml source files contain a reference to this file in an
 Emacs setup block at the bottom of the file...

25 years agoInformation on the CVS tree and remote access.
Thomas G. Lockhart [Fri, 18 Dec 1998 16:08:19 +0000 (16:08 +0000)]
Information on the CVS tree and remote access.
Not yet complete, integrated, or marked up.
Not yet included in a document (should go in the developer's doc?).

25 years agoFixed nodeToString() to put out "<>" for NULL strings again.
Jan Wieck [Fri, 18 Dec 1998 14:45:09 +0000 (14:45 +0000)]
Fixed nodeToString() to put out "<>" for NULL strings again.
More cleanups to appendStringInfo() usage in node/outfuncs.c.

Jan

25 years agoSET TRANSACTION ISOLATION LEVEL ...
Vadim B. Mikheev [Fri, 18 Dec 1998 09:10:39 +0000 (09:10 +0000)]
SET TRANSACTION ISOLATION LEVEL ...
LOCK TABLE IN ... MODE
...implemented

25 years agoMissed a few files.
Bruce Momjian [Fri, 18 Dec 1998 07:09:26 +0000 (07:09 +0000)]
Missed a few files.

25 years agoAttached is a patch with some fixes that (I think that) should go into
Bruce Momjian [Fri, 18 Dec 1998 07:08:03 +0000 (07:08 +0000)]
Attached is a patch with some fixes that (I think that) should go into
6.4.1. Here is the list:

- The type int8 now works. In fact, the bug(s) were in
src/backend/port/snprintf.c, so int8 is probably broken in every platform
that hasn't a native snprintf/vsnprintf. The type itself worked as
expected, only the output was wrong. Anyway, this patch should be checked
in other platforms.

- The regression tests for int2 and int4, which were broken due to
differences in the error messages, are fixed.

- The regression test for float8, which was broken in the reference
platform, is also fixed. I don't know if the new file (float8-OSF1.out)
will work on other platforms, but it might be worth to try it.

- Two new template files are provided (alpha_cc, which includes
optimization, and alpha_gcc), and src/templates/.similar is updated
accordingly. src/templates/alpha should be removed from the distribution.
*IMPORTANT NOTE*: I don't know if you can use gcc to compile postgres;
I've written the alpha_gcc file because alpha_cc has some flags that are
specific to DEC C.

- There is a (very basic) Digital Unix specific FAQ in
doc/FAQ_DigitalUnix.

--
-------------------------------------------------------------------
Pedro José Lobo Perea                   Tel:    +34 91 336 78 19

25 years agoFix for HAVE_LONG bug in snprintf.c.
Bruce Momjian [Fri, 18 Dec 1998 07:03:06 +0000 (07:03 +0000)]
Fix for HAVE_LONG bug in snprintf.c.

25 years agoFix for snprintf with long long's, define problem.
Bruce Momjian [Fri, 18 Dec 1998 06:59:39 +0000 (06:59 +0000)]
Fix for snprintf with long long's, define problem.

25 years agoRemove need for doc 'install man' in INSTALL file. install does both
Bruce Momjian [Thu, 17 Dec 1998 16:37:04 +0000 (16:37 +0000)]
Remove need for doc 'install man' in INSTALL file.  install does both
html and man.

25 years agoFrom: Keith Parks <emkxp01@mtcc.demon.co.uk>
Marc G. Fournier [Thu, 17 Dec 1998 13:09:52 +0000 (13:09 +0000)]
From: Keith Parks <emkxp01@mtcc.demon.co.uk>

Looks like :vartypmod got transmuted to %vartypmod in an editing session.

25 years agoUpgrade to Pygress 2.2.
Bruce Momjian [Thu, 17 Dec 1998 01:43:04 +0000 (01:43 +0000)]
Upgrade to Pygress 2.2.

25 years agoSerialized mode works!
Vadim B. Mikheev [Wed, 16 Dec 1998 11:53:55 +0000 (11:53 +0000)]
Serialized mode works!

25 years agoOops, sorry...meant to commit the patch from Thomas for tzn->CTZName
Marc G. Fournier [Tue, 15 Dec 1998 15:28:57 +0000 (15:28 +0000)]
Oops, sorry...meant to commit the patch from Thomas for tzn->CTZName

25 years agotzn undeclared in FreeBSD, commented out.
Vadim B. Mikheev [Tue, 15 Dec 1998 15:10:03 +0000 (15:10 +0000)]
tzn undeclared in FreeBSD, commented out.

25 years agoInitial MVCC code.
Vadim B. Mikheev [Tue, 15 Dec 1998 12:47:01 +0000 (12:47 +0000)]
Initial MVCC code.
New code for locking buffer' context.

25 years agoMore cleanups ... this is one big file *sigh*
Marc G. Fournier [Tue, 15 Dec 1998 04:00:46 +0000 (04:00 +0000)]
More cleanups ... this is one big file *sigh*

25 years agoCorrect some typos...
Marc G. Fournier [Tue, 15 Dec 1998 02:32:57 +0000 (02:32 +0000)]
Correct some typos...

25 years agoMore cleanups of appendStringInfo() function calls...
Marc G. Fournier [Tue, 15 Dec 1998 02:24:15 +0000 (02:24 +0000)]
More cleanups of appendStringInfo() function calls...

25 years agoCompilation of libpq for Win32 breaks on 6.4, because of a change that I
Bruce Momjian [Mon, 14 Dec 1998 15:05:27 +0000 (15:05 +0000)]
Compilation of libpq for Win32 breaks on 6.4, because of a change that I
missed before the release. It's simply a symbol that is undefined. This
patch defines this symbol in "win32.h", so it should have no effect on any
other platforms. It should go into 6.4.1 if possible, since compilation is
completely broken without it.

I am also attaching a patch for the "win32.mak" file - it leaves a file
behind when doing "make clean" after the library is built on Visual C++ 6.0.
This is not at all as urgent, but I don't see it breaking here, so I think
it might as well go in there too?

//Magnus

25 years agomore cleanups...of note, appendStringInfo now performs like sprintf(),
Marc G. Fournier [Mon, 14 Dec 1998 08:11:17 +0000 (08:11 +0000)]
more cleanups...of note, appendStringInfo now performs like sprintf(),
where you state a format and arguments.  the old behavior required
each appendStringInfo to have to have a sprintf() before it if any
formatting was required.

Also shortened several instances where there were multiple appendStringInfo()
calls in a row, doing nothing more then adding one more word to the String,
instead of doing them all in one call.

25 years agoMany more cleanups...
Marc G. Fournier [Mon, 14 Dec 1998 06:50:32 +0000 (06:50 +0000)]
Many more cleanups...

25 years agoInitial attempt to clean up the code...
Marc G. Fournier [Mon, 14 Dec 1998 05:19:16 +0000 (05:19 +0000)]
Initial attempt to clean up the code...

Switch sprintf() to snprintf()
Remove any/all #if 0 -or- #ifdef NOT_USED -or- #ifdef FALSE sections of
code

25 years agoOK, here is a diff for the README file in /usr/src/pgsql/contrib/spi/.
Bruce Momjian [Mon, 14 Dec 1998 05:13:39 +0000 (05:13 +0000)]
OK, here is a diff for the README file in /usr/src/pgsql/contrib/spi/.
For the 6.5 tree.

Have a great night.
Terry

25 years agoCurrent multi-byte related codes have a bug with SQL_ASCII
Bruce Momjian [Mon, 14 Dec 1998 04:59:58 +0000 (04:59 +0000)]
Current multi-byte related codes have a bug with SQL_ASCII
support. Included patches will solve it and should be applied to
both trees.  Also, it fix the problem with \c command of psql when
switching different encoding databases.

Regression tests passed.
--
Tatsuo Ishii
t-ishii@sra.co.jp

25 years agoAdd support for the CASE statement in the rewrite handling.
Thomas G. Lockhart [Mon, 14 Dec 1998 00:02:17 +0000 (00:02 +0000)]
Add support for the CASE statement in the rewrite handling.
Allows (at least some) rules and views.
Still some trouble (crashes) with target CASE columns spanning tables,
 but lots now works.

25 years agoImprove CASE statement support.
Thomas G. Lockhart [Sun, 13 Dec 1998 23:56:44 +0000 (23:56 +0000)]
Improve CASE statement support.
Try to label CASE columns for a SELECT if not specified with an AS clause.

25 years agoFix up error messages when looking up functions and operators to not
Thomas G. Lockhart [Sun, 13 Dec 1998 23:54:40 +0000 (23:54 +0000)]
Fix up error messages when looking up functions and operators to not
 make the sometimes misleading claim that more than one candidate was
 identified. Now say "Unable to identify...".

25 years agoUse the new implicit type coersion techniques for matching up types
Thomas G. Lockhart [Sun, 13 Dec 1998 23:50:58 +0000 (23:50 +0000)]
Use the new implicit type coersion techniques for matching up types
 between columns and DEFAULT clauses.

25 years agoClean up error messages associated with rules and views.
Thomas G. Lockhart [Sun, 13 Dec 1998 23:49:18 +0000 (23:49 +0000)]
Clean up error messages associated with rules and views.
Fix parser error message when an operator is not found
 to not explicitly claim that there is more than one.

25 years agoBuild pltcl.so correctly on platforms that want dependent
Tom Lane [Sun, 13 Dec 1998 23:46:49 +0000 (23:46 +0000)]
Build pltcl.so correctly on platforms that want dependent
shared libraries to be listed in the link command.

25 years agoAdd routines and synonyms to help with single-byte char type handling.
Thomas G. Lockhart [Sun, 13 Dec 1998 23:45:22 +0000 (23:45 +0000)]
Add routines and synonyms to help with single-byte char type handling.
Fix one usage of substr() which mapped to the "Oracle compatibility" funcs
 rather than the more recent (and closer to SQL92) function in varlena.c.
Add more DESC() entries for conversion functions.

25 years agoAdjust handling of data type names to suppress double quotes
Thomas G. Lockhart [Sun, 13 Dec 1998 23:41:32 +0000 (23:41 +0000)]
Adjust handling of data type names to suppress double quotes
 unless necessary.
Label internal bpchar types as "character" and varchar types as
 "character varying" to be less Postgres-specific. These types map to
 the SQL92 definitions anyway.
Redefine g_force_quotes to be the local variable force_quotes.
Pass this as an argument to fmtId().
These should help with handling the single-byte internal "char" type.

25 years agoAdd routines to help with single-byte (internal) character type support.
Thomas G. Lockhart [Sun, 13 Dec 1998 23:36:48 +0000 (23:36 +0000)]
Add routines to help with single-byte (internal) character type support.

25 years agoSwitch around conditional code so that HAVE_TM_ZONE takes precedence
Thomas G. Lockhart [Sun, 13 Dec 1998 23:34:17 +0000 (23:34 +0000)]
Switch around conditional code so that HAVE_TM_ZONE takes precedence
 over HAVE_INT_TIMEZONE. This may help out linux/glibc2 and Dec Alpha.
Included #error precompiler macros to catch cases where neither is defined
 but USE_POSIX_TIME is (shouldn't happen). Hopefully this isn't just
 a gcc-ism.

25 years agoPartial fix for vendor-cc-versus-gcc incompatibility on HPUX:
Tom Lane [Sun, 13 Dec 1998 20:05:16 +0000 (20:05 +0000)]
Partial fix for vendor-cc-versus-gcc incompatibility on HPUX:
prefer aCC as c++ compiler if hpux_cc is selected as template.
Doesn't solve the problem if you have g++ and not aCC, however...

25 years agoUse standard AC_PROG_INSTALL macro to search for install program,
Tom Lane [Sun, 13 Dec 1998 20:03:07 +0000 (20:03 +0000)]
Use standard AC_PROG_INSTALL macro to search for install program,
instead of our own halfway-there code.  Add AC_STRUCT_TIMEZONE call
to check whether tm_zone exists in struct tm.  Revise reading of template
file so that templates can define any variables they feel like (and,
indeed, can execute arbitrary shell code) rather than being constrained
to a fixed set of variable names.