]> granicus.if.org Git - postgresql/log
postgresql
26 years agoAdd new subselect test.
Thomas G. Lockhart [Wed, 18 Feb 1998 07:32:17 +0000 (07:32 +0000)]
Add new subselect test.

26 years agoNew table from new subselect test.
Thomas G. Lockhart [Wed, 18 Feb 1998 07:29:45 +0000 (07:29 +0000)]
New table from new subselect test.

26 years agoInclude products of gram.y to allow systems without bison to build Postgres.
Thomas G. Lockhart [Wed, 18 Feb 1998 07:28:08 +0000 (07:28 +0000)]
Include products of gram.y to allow systems without bison to build Postgres.
 The parser recently grew too big for some yaccs, including FreeBSD.

26 years agoAllow LIKE expression in constraint clause.
Thomas G. Lockhart [Wed, 18 Feb 1998 07:25:57 +0000 (07:25 +0000)]
Allow LIKE expression in constraint clause.
Allow USER and VALID as column names and identifiers.

26 years agoFix parsing for <identifier>-<number> which was mis-identified as a unary
Thomas G. Lockhart [Wed, 18 Feb 1998 07:23:22 +0000 (07:23 +0000)]
Fix parsing for <identifier>-<number> which was mis-identified as a unary
 minus. Example is SELECT f1-2 FROM INT4_TBL;

26 years agoFix from Vadim for simple-minded case SELECT 1 WHERE 1 IN (SELECT 1)
Thomas G. Lockhart [Wed, 18 Feb 1998 07:19:34 +0000 (07:19 +0000)]
Fix from Vadim for simple-minded case SELECT 1 WHERE 1 IN (SELECT 1)
 used in regression tests. No longer core dumps.

26 years agoRemove inclusion of the doc make. Looks like all of the man pages are
Thomas G. Lockhart [Wed, 18 Feb 1998 07:16:58 +0000 (07:16 +0000)]
Remove inclusion of the doc make. Looks like all of the man pages are
 in the usual man page source directory anyway. Source tree now builds.

26 years agoCancel last version which introduced unintended flags
Thomas G. Lockhart [Wed, 18 Feb 1998 07:15:23 +0000 (07:15 +0000)]
Cancel last version which introduced unintended flags
 for linux-alpha support.

26 years agoFix useor for (a,b...) Op ANY/ALL
Vadim B. Mikheev [Wed, 18 Feb 1998 03:26:54 +0000 (03:26 +0000)]
Fix useor for (a,b...) Op ANY/ALL

26 years agoI'm getting a SEGV error when testing ecpg using the perftest,or
Marc G. Fournier [Wed, 18 Feb 1998 01:30:42 +0000 (01:30 +0000)]
I'm getting a SEGV error when testing ecpg using the perftest,or
any other, example program.

I have tracked this down to a call to PQfinish() in ECPGfinish()
that occurs before any connection is established.

From: Keith Parks <emkxp01@mtcc.demon.co.uk>

26 years agoUpdate the preprocessor code
Marc G. Fournier [Wed, 18 Feb 1998 01:28:03 +0000 (01:28 +0000)]
Update the preprocessor code

From: Michael Meskes <meskes@topsystem.de>

26 years agoCheck for and include <getopt.h>
Marc G. Fournier [Wed, 18 Feb 1998 01:26:02 +0000 (01:26 +0000)]
Check for and include <getopt.h>

26 years agoJust another missing dummy file
Marc G. Fournier [Tue, 17 Feb 1998 02:09:17 +0000 (02:09 +0000)]
Just another missing dummy file

26 years ago2. IRIX has both HAVE_FPCLASS and HAVE_FP_CLASS, therefore, in
Marc G. Fournier [Tue, 17 Feb 1998 02:00:12 +0000 (02:00 +0000)]
2. IRIX has both HAVE_FPCLASS and HAVE_FP_CLASS, therefore, in
backend/port/isinf.c it tries to compile two versions of
isinf().

From: Andrew Martin <martin@biochemistry.ucl.ac.uk>

26 years agoMajor update of ecpg preprocessor
Marc G. Fournier [Tue, 17 Feb 1998 01:48:12 +0000 (01:48 +0000)]
Major update of ecpg preprocessor

From: Michael Meskes <meskes@topsystem.de>

26 years agoTotally forgot to add these files...
Marc G. Fournier [Tue, 17 Feb 1998 01:43:04 +0000 (01:43 +0000)]
Totally forgot to add these files...

26 years agoThe problem is that $ac_cv_prog_gcc is empty, instead of 'no' or
Marc G. Fournier [Tue, 17 Feb 1998 01:32:37 +0000 (01:32 +0000)]
The problem is that $ac_cv_prog_gcc is empty, instead of 'no' or
whatsoever. The patch is not a solution, because configure is generated
from configure.in, and I don't know how to patch it to get a working
'configure'.

From: "Pedro J. Lobo" <pjlobo@euitt.upm.es>

26 years agoRemove all backend/port/${PORTNAME} references from template files, since
Marc G. Fournier [Mon, 16 Feb 1998 14:39:29 +0000 (14:39 +0000)]
Remove all backend/port/${PORTNAME} references from template files, since
they don't exist anymore

26 years agoFrom: Brian E Gallew <geek+@cmu.edu>
Marc G. Fournier [Sat, 14 Feb 1998 19:56:21 +0000 (19:56 +0000)]
From: Brian E Gallew <geek+@cmu.edu>

dgux 5.4R4.11

Missing port-protos.h (not needed, I think).  Wants dld.h.  Should
really use the system dl stuff (like i386_solaris).  Needs to include
<netinet/in.h> before <arpa/inet.h>.  Here are some patches...

26 years agoApparently there was a
Marc G. Fournier [Sat, 14 Feb 1998 18:06:20 +0000 (18:06 +0000)]
Apparently there was a
compiler define that should have been enabled, but was not due to
different naming conventions for Linux/Alpha. Attached is the patch he
sent me, that I have not had a chance to test yet.

From: Ryan Kirkpatrick <rkirkpat@nag.cs.colorado.edu>

26 years agoFrom: plh@opim.uconn.edu
Marc G. Fournier [Sat, 14 Feb 1998 18:04:39 +0000 (18:04 +0000)]
From: plh@opim.uconn.edu

Missing = in an update

26 years agoRemove HBA related switch
Marc G. Fournier [Sat, 14 Feb 1998 18:02:43 +0000 (18:02 +0000)]
Remove HBA related switch

From: Tom I Helbekkmo <tih@Hamartun.Priv.NO>

26 years agoFrom: Tom I Helbekkmo <tih@Hamartun.Priv.NO>
Marc G. Fournier [Sat, 14 Feb 1998 18:00:37 +0000 (18:00 +0000)]
From: Tom I Helbekkmo <tih@Hamartun.Priv.NO>

The file 'backend/utils/adt/arrayfuncs.c' won't compile with the
February 14th snapshot, because of an inconsistency between the
declaration and implementation of ReadArrayStr().  As far as I can
tell, the predeclaration is wrong.  I assume this is what was meant:

26 years agoI'm including an update to my user defined IP and MAC address type
Marc G. Fournier [Sat, 14 Feb 1998 17:58:09 +0000 (17:58 +0000)]
I'm including an update to my user defined IP and MAC address type
implementation that's in contrib/ip_and_mac/.  This one works right
with 6.3, avoids the problems I ran into earlier with LIKE, and
includes a bit of extra functionality.

From: Tom I Helbekkmo <tih@Hamartun.Priv.NO>

26 years agoMan page for ecpg.1
Marc G. Fournier [Fri, 13 Feb 1998 19:49:06 +0000 (19:49 +0000)]
Man page for ecpg.1

26 years agoAtttypmod cleanup.
Bruce Momjian [Fri, 13 Feb 1998 19:46:22 +0000 (19:46 +0000)]
Atttypmod cleanup.

26 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Fri, 13 Feb 1998 17:18:25 +0000 (17:18 +0000)]
From: Jan Wieck <jwieck@debis.com>

    just a little correction in the pltcl_guide.nr.

    Sometimes I changed the name of tuple arguments to numbers
    like the other args are. Otherwise it wasn't possible to
    create a function as

        CREATE FUNCTION f (EMP, EMP) ... LANGUAGE 'pltcl';

    The arguments are now accessed in the function as

        $1(name) vs. $2(name)

26 years agoFrom: Zeugswetter Andreas SARZ <Andreas.Zeugswetter@telecom.at>
Marc G. Fournier [Fri, 13 Feb 1998 17:12:08 +0000 (17:12 +0000)]
From: Zeugswetter Andreas SARZ <Andreas.Zeugswetter@telecom.at>

Only occurrs in
src/include/storage/s_lock.h:#if defined(__AIX)
src/include/utils/dt.h:#if defined(__AIX)
src/include/utils/nabstime.h:#if defined(__AIX)

Simply delete one underscore, only occurs once per file, so no patch.

26 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Fri, 13 Feb 1998 13:23:33 +0000 (13:23 +0000)]
From: Jan Wieck <jwieck@debis.com>

    Someone changed the parser to build a TypeName node on CREATE
    FUNCTION in any  case.  As  a  side  effect,  ALL!  functions
    created  got  the  proretset  attribute  to  true. Thus for a
    SELECT the parser wrapped an Iter node around  the  Expr  and
    since  singleton  functions  set  isDone  the Iter returns no
    tuple up.

26 years agoAdd missing extern.h file
Marc G. Fournier [Fri, 13 Feb 1998 13:15:23 +0000 (13:15 +0000)]
Add missing extern.h file

26 years agoMakefile.in needs to be configured to use PORTNAME variable
Marc G. Fournier [Fri, 13 Feb 1998 13:15:12 +0000 (13:15 +0000)]
Makefile.in needs to be configured to use PORTNAME variable

26 years agoRemove interfaces/ecpg/{include,preproc}/Makefile.in from configure
Marc G. Fournier [Fri, 13 Feb 1998 13:14:41 +0000 (13:14 +0000)]
Remove interfaces/ecpg/{include,preproc}/Makefile.in from configure

26 years agogram.y: ALL_SUBLINK type was returned for x Op (subquery).
Vadim B. Mikheev [Fri, 13 Feb 1998 08:10:33 +0000 (08:10 +0000)]
gram.y: ALL_SUBLINK type was returned for x Op (subquery).
parse_expr.c: only Op of bool type are supported currently...

26 years agoFrom: Tom I Helbekkmo <tih@Hamartun.Priv.NO>
Marc G. Fournier [Fri, 13 Feb 1998 05:10:06 +0000 (05:10 +0000)]
From: Tom I Helbekkmo <tih@Hamartun.Priv.NO>

Apart from this Makefile hack, all I've done is to make dynamically
loaded code modules fail properly (as was already done for __mips__,
although I think this is too loose: I believe NetBSD for the pmax can
do dynamic linking), and to add test-and-set lock handling.  As Bruce
suggested, this is done in a maximally efficient inlined way: I was
not aware that this code was so important, speed-wise.

26 years agoRemove -m486 option, since it is platform specific...this should technically
Marc G. Fournier [Fri, 13 Feb 1998 04:57:23 +0000 (04:57 +0000)]
Remove -m486 option, since it is platform specific...this should technically
be removed everywhere, but so far linux-elf is the only platform that is
being reported a problem with...

26 years agoSubselects...
Vadim B. Mikheev [Fri, 13 Feb 1998 03:46:56 +0000 (03:46 +0000)]
Subselects...

26 years agoNew SubPlan node for subselects.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:45:29 +0000 (03:45 +0000)]
New SubPlan node for subselects.
New PARAM_EXEC type.

26 years agoEState->->es_param_exec_vals = NULL; in CreateExecutorState().
Vadim B. Mikheev [Fri, 13 Feb 1998 03:42:32 +0000 (03:42 +0000)]
EState->->es_param_exec_vals = NULL; in CreateExecutorState().

26 years agoFix for EXISTS.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:41:23 +0000 (03:41 +0000)]
Fix for EXISTS.

26 years agoSupport for subselects.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:40:23 +0000 (03:40 +0000)]
Support for subselects.

26 years agoCall union_planner() instead of planner().
Vadim B. Mikheev [Fri, 13 Feb 1998 03:39:26 +0000 (03:39 +0000)]
Call union_planner() instead of planner().

26 years agoOld planner() becomes union_planner(); new planner() makes initialization
Vadim B. Mikheev [Fri, 13 Feb 1998 03:37:04 +0000 (03:37 +0000)]
Old planner() becomes union_planner(); new planner() makes initialization
of some global variables to support subselects and calls union_planner().
Calls to SS_replace_correlation_vars() and SS_process_sublinks() in
query_planner() before planning.
Get rid of #ifdef INDEXSCAN_PATCH in createplan.c.

26 years agoQuite limited support for subselects in clausesel.c.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:29:39 +0000 (03:29 +0000)]
Quite limited support for subselects in clausesel.c.
Get rid of #ifdef INDEXSCAN_PATCH in indxpath.c.

26 years agoSupport for subselects.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:27:47 +0000 (03:27 +0000)]
Support for subselects.
(Have to re-visit readfuncs.c)

26 years agoSupport for subselects.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:26:53 +0000 (03:26 +0000)]
Support for subselects.
ExecReScan for nodeAgg, nodeHash, nodeHashjoin, nodeNestloop and nodeResult.
Fixed ExecReScan for nodeMaterial.
Get rid of #ifdef INDEXSCAN_PATCH.
Get rid of ExecMarkPos and ExecRestrPos in nodeNestloop.

26 years agoSupport for subselects.
Vadim B. Mikheev [Fri, 13 Feb 1998 03:21:30 +0000 (03:21 +0000)]
Support for subselects.

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Thu, 12 Feb 1998 14:02:10 +0000 (14:02 +0000)]
From: Michael Meskes <meskes@topsystem.de>

Here's the ecpg patch for the local variables bug I reported earlier:

26 years agoThere, fixed
Marc G. Fournier [Thu, 12 Feb 1998 12:38:12 +0000 (12:38 +0000)]
There, fixed

26 years agoScrewed this up...fixing now
Marc G. Fournier [Thu, 12 Feb 1998 12:37:49 +0000 (12:37 +0000)]
Screwed this up...fixing now

26 years agoClean up Makefiles - shouldn't require configure *.in Makefiles, as they should
Marc G. Fournier [Thu, 12 Feb 1998 12:35:56 +0000 (12:35 +0000)]
Clean up Makefiles - shouldn't require configure *.in Makefiles, as they should
pull their data/info from Makefile.global

26 years agoRemove configure code from ecpg, as well as remove the 'for' loop from the
Marc G. Fournier [Thu, 12 Feb 1998 02:14:59 +0000 (02:14 +0000)]
Remove configure code from ecpg, as well as remove the 'for' loop from the
Makefile, as it isn't appropriate for GNUmake

26 years agoMemSet cleanup.
Bruce Momjian [Thu, 12 Feb 1998 01:50:01 +0000 (01:50 +0000)]
MemSet cleanup.

26 years agoMemset cleanup
Bruce Momjian [Wed, 11 Feb 1998 21:45:40 +0000 (21:45 +0000)]
Memset cleanup

26 years agoChange to MemSet for Alpha.
Bruce Momjian [Wed, 11 Feb 1998 21:38:08 +0000 (21:38 +0000)]
Change to MemSet for Alpha.

26 years agoChange int align.
Bruce Momjian [Wed, 11 Feb 1998 21:17:44 +0000 (21:17 +0000)]
Change int align.

26 years agoGoodbye register keyword. Compiler knows better.
Bruce Momjian [Wed, 11 Feb 1998 19:14:04 +0000 (19:14 +0000)]
Goodbye register keyword.  Compiler knows better.

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Wed, 11 Feb 1998 15:30:00 +0000 (15:30 +0000)]
From: Michael Meskes <meskes@topsystem.de>

Cleanups for ecpg, as well as a missing patch so that its configured in

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Wed, 11 Feb 1998 15:18:05 +0000 (15:18 +0000)]
From: Michael Meskes <meskes@topsystem.de>

Cleanups for ecpg, as well as a missing patch so that its configured in

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Wed, 11 Feb 1998 14:44:02 +0000 (14:44 +0000)]
From: Michael Meskes <meskes@topsystem.de>

Remove "temp" configure files

26 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Wed, 11 Feb 1998 14:38:31 +0000 (14:38 +0000)]
From: Jan Wieck <jwieck@debis.com>

    A few minutes ago I sent down the PL/Tcl  directory  to  this
    list.  Look at it and reuse anything that might help to build
    PL/perl.  I really hope that PL/perl and PL/Tcl appear in the
    6.3 distribution. I'll do whatever I can to make this happen.

26 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Wed, 11 Feb 1998 14:14:18 +0000 (14:14 +0000)]
From: Jan Wieck <jwieck@debis.com>

    A few minutes ago I sent down the PL/Tcl  directory  to  this
    list.  Look at it and reuse anything that might help to build
    PL/perl.  I really hope that PL/perl and PL/Tcl appear in the
    6.3 distribution. I'll do whatever I can to make this happen.

26 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Wed, 11 Feb 1998 14:08:01 +0000 (14:08 +0000)]
From: Jan Wieck <jwieck@debis.com>

    A few minutes ago I sent down the PL/Tcl  directory  to  this
    list.  Look at it and reuse anything that might help to build
    PL/perl.  I really hope that PL/perl and PL/Tcl appear in the
    6.3 distribution. I'll do whatever I can to make this happen.

26 years agoRemove some definitions which could not work; conversions from datetime
Thomas G. Lockhart [Wed, 11 Feb 1998 04:24:20 +0000 (04:24 +0000)]
Remove some definitions which could not work; conversions from datetime
 to integer unix system time conflict on the input types.
 Leave in the conversions from integer unix system time to datetime.

26 years agoUpdate test for newly implemented "<=" operator.
Thomas G. Lockhart [Wed, 11 Feb 1998 04:13:33 +0000 (04:13 +0000)]
Update test for newly implemented "<=" operator.

26 years agoTry bumping to version two...
Thomas G. Lockhart [Wed, 11 Feb 1998 04:11:19 +0000 (04:11 +0000)]
Try bumping to version two...

26 years agoDefine ROW and STATEMENT as parser tokens.
Thomas G. Lockhart [Wed, 11 Feb 1998 04:09:54 +0000 (04:09 +0000)]
Define ROW and STATEMENT as parser tokens.
Use explicit tokens to decode CREATE TRIGGER clauses.
Allow ROW and STATEMENT as column identifiers.
Fix CAST syntax to require parens per SQL92 spec.
Define TypeId to allow correct translation of type names in CREATE FUNCTION
 and other statements. Need to do this without looking up defined type
 names because CREATE FUNCTION can specify undefined (new) types.
Define UserId to complete removal of "Id" generic entity.
Define xlateSqlFunc() to convert SQL92 CHARACTER_LENGTH() and CHAR_LENGTH()
 functions to calls to length().
Define func_name parser entity for contexts requiring a function name.
Have xlateSqlType() translate "float" to "float8".

26 years agoAutomatically promote out of range integers to floats.
Thomas G. Lockhart [Wed, 11 Feb 1998 03:56:08 +0000 (03:56 +0000)]
Automatically promote out of range integers to floats.
Throw elog(NOTICE) to flag promotion.

26 years agoFix typos in SQL statement terminators.
Thomas G. Lockhart [Wed, 11 Feb 1998 03:51:38 +0000 (03:51 +0000)]
Fix typos in SQL statement terminators.

26 years agoregression test based off of Solaris/Sparc v2.6
Marc G. Fournier [Tue, 10 Feb 1998 17:42:54 +0000 (17:42 +0000)]
regression test based off of Solaris/Sparc v2.6

26 years agoErk, the whole directory structure changed on us here...
Marc G. Fournier [Tue, 10 Feb 1998 16:44:17 +0000 (16:44 +0000)]
Erk, the whole directory structure changed on us here...

26 years agoUpdated ECPG From: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Tue, 10 Feb 1998 16:37:01 +0000 (16:37 +0000)]
Updated ECPG From: Michael Meskes <meskes@topsystem.de>

26 years agoPass around typmod as int16.
Bruce Momjian [Tue, 10 Feb 1998 16:04:38 +0000 (16:04 +0000)]
Pass around typmod as int16.

26 years agoClean up regression tests for SunOS (based on Solaris v2.6)
Marc G. Fournier [Tue, 10 Feb 1998 14:22:50 +0000 (14:22 +0000)]
Clean up regression tests for SunOS (based on Solaris v2.6)

Clean up strings.out , removed func_get_detail from error message

26 years agoPass attypmod through to executor by adding to Var and Resdom.
Bruce Momjian [Tue, 10 Feb 1998 04:02:59 +0000 (04:02 +0000)]
Pass attypmod  through to executor by adding to Var and Resdom.

26 years agoFrom: Peter T Mount <patches@maidast.demon.co.uk>
Marc G. Fournier [Mon, 9 Feb 1998 03:22:41 +0000 (03:22 +0000)]
From: Peter T Mount <patches@maidast.demon.co.uk>

This patch fixes the following:

* Fixes minor bug found in DatabaseMetaData.getTables() where it doesn't
  handle default table types.
* It now reports an error if the client opens a database using
  properties, and either the user or password properties are missing. This
  should make the recent problem with Servlets easier to find.
* Commented out obsolete property in Driver.getPropertyInfo()

26 years agoThis was a new class that was introduced last weekend. I'm attaching the
Marc G. Fournier [Mon, 9 Feb 1998 02:16:55 +0000 (02:16 +0000)]
This was a new class that was introduced last weekend. I'm attaching the
source here. It goes in the src/interfaces/jdbc/postgresql/util directory

Marc, can you make sure this is in the distribution, Thanks

26 years agoAtttypmod cleanups.
Bruce Momjian [Sat, 7 Feb 1998 21:41:52 +0000 (21:41 +0000)]
Atttypmod cleanups.

26 years agoatttypmod now -1.
Bruce Momjian [Sat, 7 Feb 1998 06:11:56 +0000 (06:11 +0000)]
atttypmod now -1.

26 years agoUpdate now that attcacheoff initial value is -1 always.
Bruce Momjian [Fri, 6 Feb 1998 20:18:01 +0000 (20:18 +0000)]
Update now that attcacheoff initial value is -1 always.

26 years agoSet attcacheoff value to -1 for all tables.
Bruce Momjian [Fri, 6 Feb 1998 19:18:06 +0000 (19:18 +0000)]
Set attcacheoff value to -1 for all tables.

26 years agoFix for \d on long table names.
Bruce Momjian [Fri, 6 Feb 1998 17:46:31 +0000 (17:46 +0000)]
Fix for \d on long table names.

26 years agoFIx hasSubLinks for Vadim.
Bruce Momjian [Fri, 6 Feb 1998 16:46:29 +0000 (16:46 +0000)]
FIx hasSubLinks for Vadim.

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Fri, 6 Feb 1998 13:32:34 +0000 (13:32 +0000)]
From: Michael Meskes <meskes@topsystem.de>

I have implemented a better user interface (well part of) so you can use it
as expected. As usual there are some bug fixes. :-)

26 years agoFIx for varchar(), char() and INSERT .. SELECT.
Bruce Momjian [Thu, 5 Feb 1998 22:48:44 +0000 (22:48 +0000)]
FIx for varchar(), char() and INSERT .. SELECT.

26 years agoVadim fix for vacuum analyze.
Bruce Momjian [Thu, 5 Feb 1998 21:19:21 +0000 (21:19 +0000)]
Vadim fix for vacuum analyze.

26 years agoFIx for atttypmod on system tables.
Bruce Momjian [Thu, 5 Feb 1998 19:50:55 +0000 (19:50 +0000)]
FIx for atttypmod on system tables.

26 years agoFIx for atttypmod in system catalogs.
Bruce Momjian [Thu, 5 Feb 1998 19:02:44 +0000 (19:02 +0000)]
FIx for atttypmod in system catalogs.

26 years agoFix for varchar functions, and indextyple j-1 fix.
Bruce Momjian [Thu, 5 Feb 1998 17:22:41 +0000 (17:22 +0000)]
Fix for varchar functions, and indextyple j-1 fix.

26 years agoFrom: Michael Meskes <meskes@topsystem.de>
Marc G. Fournier [Thu, 5 Feb 1998 15:46:43 +0000 (15:46 +0000)]
From: Michael Meskes <meskes@topsystem.de>

Well this is not really a patch. But I mananged to get Linus' old Postgres95
precompiler to compile and work with PostgreSQL. The next step would be to
collect bug/missing feature reports and to put it into the distribution so
that it is made with the standard make procedure.

Warning! So far it is not tested much and it does not install correctly. But
I was able to create a small binary with it.

26 years ago Symptom:
Marc G. Fournier [Thu, 5 Feb 1998 15:08:49 +0000 (15:08 +0000)]
Symptom:
        select  from  a  table  with  attrs  (a  int, b char(20))
        crashed in bpcharout() (palloc of -1 bytes). But a  table
        with attrs (a int, b varchar(20)) worked.

From: Jan Wieck <jwieck@debis.com>

26 years agoinit_fixup_addr() function removed, so port-protos.h no longer required
Marc G. Fournier [Thu, 5 Feb 1998 04:21:56 +0000 (04:21 +0000)]
init_fixup_addr() function removed, so port-protos.h no longer required

26 years agofunc_error() changed so that if caller is passed with NULL value, its
Marc G. Fournier [Thu, 5 Feb 1998 04:08:44 +0000 (04:08 +0000)]
func_error() changed so that if caller is passed with NULL value, its
output at least doesn't appear that its missing something.

wasn't particularly confident with removing 'caller' altogether :(

26 years agoBuild in a 'dummy' tas file...nothing in it, just used for the link process
Marc G. Fournier [Thu, 5 Feb 1998 04:00:55 +0000 (04:00 +0000)]
Build in a 'dummy' tas file...nothing in it, just used for the link process
of configure

26 years agochar() cleanup and more cacheoff improvements.
Bruce Momjian [Thu, 5 Feb 1998 03:47:08 +0000 (03:47 +0000)]
char() cleanup and more cacheoff improvements.

26 years agoFix up this so that it also doesn't use a caller variable...
Marc G. Fournier [Thu, 5 Feb 1998 03:40:10 +0000 (03:40 +0000)]
Fix up this so that it also doesn't use a caller variable...

26 years agoClean up func_error() function completely removing caller variable as
Marc G. Fournier [Thu, 5 Feb 1998 03:35:48 +0000 (03:35 +0000)]
Clean up func_error() function completely removing caller variable as
"more info then user requires"

26 years agoPort specific, compiler specfied for SGI/Irix5
Marc G. Fournier [Thu, 5 Feb 1998 03:31:01 +0000 (03:31 +0000)]
Port specific, compiler specfied for SGI/Irix5

From: Andrew Martin <martin@biochemistry.ucl.ac.uk>

26 years agoCleanup getattr code. Make CHAR() use attcacheoff.
Bruce Momjian [Wed, 4 Feb 1998 21:32:12 +0000 (21:32 +0000)]
Cleanup getattr code.  Make CHAR() use attcacheoff.

26 years agoTry this fix for the tas.s code...
Marc G. Fournier [Wed, 4 Feb 1998 13:19:32 +0000 (13:19 +0000)]
Try this fix for the tas.s code...

26 years agoAllow most expressions in BETWEEN clause. Allow all expressions if
Thomas G. Lockhart [Wed, 4 Feb 1998 06:11:48 +0000 (06:11 +0000)]
Allow most expressions in BETWEEN clause. Allow all expressions if
 surrounded by parentheses (but not all are meaningful).
Remove unused keywords ACL, APPEND, MERGE.
Requires a "make clean" to recompile all code since keyword numeric
 assignments have changed with keyword removal.