]> granicus.if.org Git - postgresql/log
postgresql
25 years agoFix for select bug.
Bruce Momjian [Wed, 26 Aug 1998 03:17:32 +0000 (03:17 +0000)]
Fix for select bug.

25 years agocvs add'd two files for the tprintf() patch...
Marc G. Fournier [Tue, 25 Aug 1998 21:43:47 +0000 (21:43 +0000)]
cvs add'd two files for the tprintf() patch...

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:37:08 +0000 (21:37 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

>       these patches define the UNLISTEN sql command. The code already
>       existed but it was unknown to the parser. Now it can be used
>       like the listen command.
>       You must make clean and delete gram.c and parser.h before make.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:34:10 +0000 (21:34 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

> tprintf.patch
>
>       tprintf.patch
>
>       adds functions and macros which implement a conditional trace package
>       with the ability to change flags and numeric options of running
>       backends at runtime.
>       Options/flags can be specified in the command line and/or read from
>       the file pg_options in the data directory.

25 years agoMassimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:32:10 +0000 (21:32 +0000)]
Massimo Dal Zotto <dz@cs.unitn.it>

> socket-flock.patch
>
>       use advisory locks to check if the unix socket can be deleted.
>       A running postmaster keeps a lock on that file. A starting
>       postmaster exits if the file exists and is locked, otherwise
>       it deletes the sockets and proceeds.
>       This avoid the need to remove manually the file after a postmaster
>       or system crash.
>       I don't know if flock is available on any system. If not we could
>       define a HAVE_FLOCK set by configure.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:31:20 +0000 (21:31 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

> sinval.patch
>
>       fixes a problem in SI cache which causes table overflow if some
>       backend is idle for a long time while other backends keep adding
>       entries.
>       It uses the new signal handling implemented in tprintf.patch.
>       I have also increacasesed the max number of backends from 32 to 64
>       and the table size from 1000 to 5000.
>       I don't know if anybody is working on SI, but until another
>       solution is found this patch fixes the problem. I have received
>       messages from other people reporting the same problem which I
>       fixed many months ago.

25 years agomissed adding a new include file
Marc G. Fournier [Tue, 25 Aug 1998 21:29:16 +0000 (21:29 +0000)]
missed adding a new include file

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:25:46 +0000 (21:25 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

> sequence.patch
>
>       adds the missing setval command to sequences. Owner of sequences
>       can now set the last value to any value between min and max
>       without recreating the sequence. This is useful after loading
>       data from external files.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:24:12 +0000 (21:24 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

> ps-status.patch
>
>       macros for ps status, used by postgres.c and utility.c.
>       Unfortunately ps status is system dependent and the current
>       code doesn't work on linux. The use of macros confines system
>       dependency to into one file (ps-status.h). Users of other
>       operating systems should check this code and submit new macros.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:22:27 +0000 (21:22 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

> pqpacket.patch
>
>       fixed indentation.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:20:32 +0000 (21:20 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

lock.patch

        I have rewritten lock.c cleaning up the code and adding better
        assert checking I have also added some fields to the lock and
        xid tags for better support of user locks. There is also a new
        function which returns an array of pids owning a lock.
        I'm using this code from over six months and it works fine.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:17:11 +0000 (21:17 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

config.patch

        remove old defines no more used in the code.

25 years agoFrom: Massimo Dal Zotto <dz@cs.unitn.it>
Marc G. Fournier [Tue, 25 Aug 1998 21:04:41 +0000 (21:04 +0000)]
From: Massimo Dal Zotto <dz@cs.unitn.it>

assert.patch

        adds a switch to turn on/off the assert checking if enabled at compile
        time. You can now compile postgres with assert checking and disable it
        at runtime in a production environment.

25 years agoCan someone please apply this portability patch to genbki.sh ?
Bruce Momjian [Tue, 25 Aug 1998 17:36:18 +0000 (17:36 +0000)]
Can someone please apply this portability patch to genbki.sh ?
(Mark or Bruce?) It fixes a problem when cpp gives a warning when
precompiling /dev/null like:  "/dev/null", line 1: 1506-229 (W)
File is empty.  This leads to a hangup when doing the description
load during initdb, since stderr also ends up in the global1.description
and local1_template1.description

stderr has to be redirected to /dev/null:

Andreas Zeugswetter

25 years agoAdd is_sequence flag to ColumnDef structure. Used to implement SERIAL type.
Thomas G. Lockhart [Tue, 25 Aug 1998 15:09:31 +0000 (15:09 +0000)]
Add is_sequence flag to ColumnDef structure. Used to implement SERIAL type.

25 years agoSupport SERIAL column type. Expand column marked is_sequence into three
Thomas G. Lockhart [Tue, 25 Aug 1998 15:08:12 +0000 (15:08 +0000)]
Support SERIAL column type. Expand column marked is_sequence into three
 statements:
 - the table definition with a default clause referencing the sequence;
 - a CREATE SEQUENCE statement;
 - a UNIQUE constraint, which expands into a CREATE INDEX statement.
This is not a perfect solution, since the sequence will remain even if
 the table is dropped. Also, there is no absolute protection on updating
 the sequence column.

25 years agoSupport SERIAL column type. Expand into an integer column but mark
Thomas G. Lockhart [Tue, 25 Aug 1998 15:04:24 +0000 (15:04 +0000)]
Support SERIAL column type. Expand into an integer column but mark
 is_sequence in the ColumnDef structure.

25 years agoMake new section header.
Thomas G. Lockhart [Tue, 25 Aug 1998 15:02:54 +0000 (15:02 +0000)]
Make new section header.

25 years agoClean up double-quotes around column and table names.
Thomas G. Lockhart [Tue, 25 Aug 1998 15:02:04 +0000 (15:02 +0000)]
Clean up double-quotes around column and table names.
 Had too many in some cases because there was already a routine
 taking care of it.

25 years agoMove debugging printout of the query tree to print for all cases.
Thomas G. Lockhart [Tue, 25 Aug 1998 15:00:17 +0000 (15:00 +0000)]
Move debugging printout of the query tree to print for all cases.
 Formerly came just after early exit from loop for command nodes,
 so missed some cases.

25 years agoAdd nextstep to similar template file
Marc G. Fournier [Tue, 25 Aug 1998 14:12:27 +0000 (14:12 +0000)]
Add nextstep to similar template file

25 years agore-integrate nextstep dynloader functionality
Marc G. Fournier [Tue, 25 Aug 1998 14:07:13 +0000 (14:07 +0000)]
re-integrate nextstep dynloader functionality

From: Jacek Lasecki <jacek@sound.eti.pg.gda.pl>

25 years agoFrom: Michael Meskes <meskes@online-club.de>
Marc G. Fournier [Tue, 25 Aug 1998 14:03:29 +0000 (14:03 +0000)]
From: Michael Meskes <meskes@online-club.de>

+
+ Fri Aug 14 12:44:21 CEST 1998
+
+       - Added EXEC SQL DEFINE statement
+       - Set version to 2.4.0
+
+ Tue Aug 18 09:24:15 CEST 1998
+
+       - Removed keyword IS from DEFINE statement
+       - Added latest changes from gram.y
+       - Removed duplicate symbols from preproc.y
+       - Initialize sqlca structure
+       - Added check for connection to ecpglib
+       - Set version to 2.4.1
+
+ Thu Aug 20 15:31:29 CEST 1998
+
+       - Cleaned up memory allocation in ecpglib.c
+       - Set library version to 2.6
+

25 years agoFrom: Michael Meskes <meskes@online-club.de>
Marc G. Fournier [Tue, 25 Aug 1998 12:17:27 +0000 (12:17 +0000)]
From: Michael Meskes <meskes@online-club.de>

+
+ Fri Aug 14 12:44:21 CEST 1998
+
+       - Added EXEC SQL DEFINE statement
+       - Set version to 2.4.0
+
+ Tue Aug 18 09:24:15 CEST 1998
+
+       - Removed keyword IS from DEFINE statement
+       - Added latest changes from gram.y
+       - Removed duplicate symbols from preproc.y
+       - Initialize sqlca structure
+       - Added check for connection to ecpglib
+       - Set version to 2.4.1
+
+ Thu Aug 20 15:31:29 CEST 1998
+
+       - Cleaned up memory allocation in ecpglib.c
+       - Set library version to 2.6
+

25 years agoFrom: Michael Meskes <meskes@online-club.de>
Marc G. Fournier [Tue, 25 Aug 1998 11:57:06 +0000 (11:57 +0000)]
From: Michael Meskes <meskes@online-club.de>

+
+ Fri Aug 14 12:44:21 CEST 1998
+
+       - Added EXEC SQL DEFINE statement
+       - Set version to 2.4.0
+
+ Tue Aug 18 09:24:15 CEST 1998
+
+       - Removed keyword IS from DEFINE statement
+       - Added latest changes from gram.y
+       - Removed duplicate symbols from preproc.y
+       - Initialize sqlca structure
+       - Added check for connection to ecpglib
+       - Set version to 2.4.1
+
+ Thu Aug 20 15:31:29 CEST 1998
+
+       - Cleaned up memory allocation in ecpglib.c
+       - Set library version to 2.6
+

25 years ago>Applied.
Bruce Momjian [Tue, 25 Aug 1998 04:19:16 +0000 (04:19 +0000)]
>Applied.

Thanks. But patches for src/backend/catalog/Makefile seems missing
in the current source tree. Please apply attached patches.

It also includes some corrections to src/backend/util/mb/wchar.c.
-- Tatsuo Ishii t-ishii@sra.co.jp

25 years agoCleanup of target file.
Bruce Momjian [Tue, 25 Aug 1998 03:22:49 +0000 (03:22 +0000)]
Cleanup of target file.

25 years agoMake sure resdomno for update/insert match attribute number for
Bruce Momjian [Tue, 25 Aug 1998 03:17:29 +0000 (03:17 +0000)]
Make sure resdomno for update/insert match attribute number for
rewrite system.  Restructure parse_target to make it easier to
understand.

25 years agoint64 fix
Bruce Momjian [Tue, 25 Aug 1998 02:23:58 +0000 (02:23 +0000)]
int64 fix

25 years agomove plpgsql
Bruce Momjian [Mon, 24 Aug 1998 19:16:28 +0000 (19:16 +0000)]
move plpgsql

25 years agomove plpgsql to src/pl for Jan.
Bruce Momjian [Mon, 24 Aug 1998 19:14:51 +0000 (19:14 +0000)]
move plpgsql to src/pl for Jan.

25 years agoFix bootstrap so it properly defines alignment of attributes.
Bruce Momjian [Mon, 24 Aug 1998 19:04:04 +0000 (19:04 +0000)]
Fix bootstrap so it properly defines alignment of attributes.

25 years agoautoconf
Bruce Momjian [Mon, 24 Aug 1998 15:43:50 +0000 (15:43 +0000)]
autoconf

25 years agoLONG_LONG_64 fix.
Bruce Momjian [Mon, 24 Aug 1998 15:38:13 +0000 (15:38 +0000)]
LONG_LONG_64 fix.

25 years agoautoconf
Bruce Momjian [Mon, 24 Aug 1998 04:17:51 +0000 (04:17 +0000)]
autoconf

25 years agoI have found a minor problem with current configure.in.
Bruce Momjian [Mon, 24 Aug 1998 04:09:39 +0000 (04:09 +0000)]
I have found a minor problem with current configure.in.

[AC_MSG_RESULT(yes) AC_DEFINE(HAVE_LONG_INT_64)],

this line produces something like:

  echo "$ac_t""yes" 1>&6 cat >> confdefs.h <<\EOF

and would append garbage "yes cat" to confdefs.h. Of course the
result confdefs.h is not syntactically correct therefore following
tests using confdefs.h would all fail.  To avoid the problem, we
could switch the order of AC_MSG_RESULT and AC_DEFINE (see attached
patch). This happend on my LinuxPPC box.

Tatsuo Ishii t-ishii@sra.co.jp

25 years agoman pages cleanup
Bruce Momjian [Mon, 24 Aug 1998 01:51:20 +0000 (01:51 +0000)]
man pages cleanup

25 years agoOLD has to return CURRENT for now, Jan.
Bruce Momjian [Mon, 24 Aug 1998 01:39:18 +0000 (01:39 +0000)]
OLD has to return CURRENT for now, Jan.

25 years ago This is the final state of the rule system for 6.4 after the
Bruce Momjian [Mon, 24 Aug 1998 01:38:11 +0000 (01:38 +0000)]
This  is the final state of the rule system for 6.4 after the
    patch is applied:

Rewrite rules on relation level work fine now.

Event qualifications on insert/update/delete  rules  work
fine now.

I  added  the  new  keyword  OLD to reference the CURRENT
tuple. CURRENT will be removed in 6.5.

Update rules can  reference  NEW  and  OLD  in  the  rule
qualification and the actions.

Insert/update/delete rules on views can be established to
let them behave like real tables.

For  insert/update/delete  rules  multiple  actions   are
supported  now.   The  actions  can also be surrounded by
parantheses to make psql  happy.   Multiple  actions  are
required if update to a view requires updates to multiple
tables.

Regular users  are  permitted  to  create/drop  rules  on
tables     they     have     RULE     permissions     for
(DefineQueryRewrite() is  now  able  to  get  around  the
access  restrictions  on  pg_rewrite).  This enables view
creation for regular users too. This  required  an  extra
boolean  parameter  to  pg_parse_and_plan() that tells to
set skipAcl on all rangetable entries  of  the  resulting
queries.       There      is      a      new     function
pg_exec_query_acl_override()  that  could  be   used   by
backend utilities to use this facility.

All rule actions (not only views) inherit the permissions
of the event relations  owner.  Sample:  User  A  creates
tables    T1    and    T2,   creates   rules   that   log
INSERT/UPDATE/DELETE on T1 in T2 (like in the  regression
tests  for rules I created) and grants ALL but RULE on T1
to user B.  User B  can  now  fully  access  T1  and  the
logging  happens  in  T2.  But user B cannot access T2 at
all, only the rule actions can. And due to  missing  RULE
permissions on T1, user B cannot disable logging.

Rules  on  the  attribute  level are disabled (they don't
work properly and since regular users are  now  permitted
to create rules I decided to disable them).

Rules  on  select  must have exactly one action that is a
select (so select rules must be a view definition).

UPDATE NEW/OLD rules  are  disabled  (still  broken,  but
triggers can do it).

There are two new system views (pg_rule and pg_view) that
show the definition of the rules or views so the db admin
can  see  what  the  users do. They use two new functions
pg_get_ruledef() and pg_get_viewdef() that are  builtins.

The functions pg_get_ruledef() and pg_get_viewdef() could
be used to implement rule and view support in pg_dump.

PostgreSQL is now the only database system I  know,  that
has rewrite rules on the query level. All others (where I
found a  rule  statement  at  all)  use  stored  database
procedures  or  the  like  (triggers as we call them) for
active rules (as some call them).

    Future of the rule system:

The now disabled parts  of  the  rule  system  (attribute
level,  multiple  actions on select and update new stuff)
require a complete new rewrite handler from scratch.  The
old one is too badly wired up.

After  6.4  I'll  start to work on a new rewrite handler,
that fully supports the attribute level  rules,  multiple
actions on select and update new.  This will be available
for 6.5 so we get full rewrite rule capabilities.

Jan

25 years agoCompile ecpg by default.
Bruce Momjian [Mon, 24 Aug 1998 01:18:51 +0000 (01:18 +0000)]
Compile ecpg by default.

25 years ago just that the regression tests for rules work, please apply
Bruce Momjian [Mon, 24 Aug 1998 01:17:46 +0000 (01:17 +0000)]
just that the regression tests for rules work, please apply
    the following to regress/sql/tests.

    If applying by hand note that the setup_... must run before
    the run_... (that I splitted these two was due to the errors
    that occured when creating rules and using them then in the
    same session - I'll post another fix for this later).

    BTW: the regression tests sanity_checks and alter_table fail
    now due to the remove of some indices and the oidint4 and
    oidname types. At least expectes should be set to the current
    results.

    Thanks.

Jan

25 years agoo note that now pg_database has a new attribuite "encoding" even
Bruce Momjian [Mon, 24 Aug 1998 01:14:24 +0000 (01:14 +0000)]
o note that now pg_database has a new attribuite "encoding" even
if MULTIBYTE is not enabled. So be sure to run initdb.

o these patches are made against the latest source tree (after
Bruce's massive patch, I think) BTW, I noticed that after running
regression, the oid field of pg_type seems disappeared.

regression=> select oid from pg_type; ERROR:  attribute
'oid' not found

this happens after the constraints test. This occures with/without
my patches. strange...

o pg_database_mb.h, pg_class_mb.h, pg_attribute_mb.h are no longer
used, and shoud be removed.

o GetDatabaseInfo() in utils/misc/database.c removed (actually in
#ifdef 0). seems nobody uses.

t-ishii@sra.co.jp

25 years agoautoconf
Bruce Momjian [Sun, 23 Aug 1998 22:36:41 +0000 (22:36 +0000)]
autoconf

25 years agoAttached is a patch that uses autoconf to determine whether there
Bruce Momjian [Sun, 23 Aug 1998 22:25:54 +0000 (22:25 +0000)]
Attached is a patch that uses autoconf to determine whether there
is a working 64-bit-int type available.

In playing around with it on my machine, I found that gcc provides
perfectly fine support for "long long" arithmetic ... but sprintf()
and sscanf(), which are system-supplied, don't work :-(.  So the
autoconf test program does a cursory test on them too.

If we find that a lot of systems are like this, it might be worth
the trouble to implement binary<->ASCII conversion of int64 ourselves
rather than relying on sprintf/sscanf to handle the data type.

regards, tom lane

25 years agocleanup
Bruce Momjian [Sun, 23 Aug 1998 14:43:46 +0000 (14:43 +0000)]
cleanup

25 years agofulltext copy script.
Bruce Momjian [Sat, 22 Aug 1998 12:54:58 +0000 (12:54 +0000)]
fulltext copy script.

25 years ago As proposed, here is the current version of PL/pgSQL. The
Bruce Momjian [Sat, 22 Aug 1998 12:38:39 +0000 (12:38 +0000)]
As  proposed,  here  is  the current version of PL/pgSQL. The
    test isn't that complete up to now,  but  I  think  it  shows
    enough of the capabilities of the module.

    The  Makefile  assumes  it  is  located  in a directory under
    pgsql/src/pl.   Since   it   includes   Makefile.global   and
    Makefile.port  and doesn't use any own compiler/linker calls,
    it should build on most of our supported  platforms  (I  only
    tested  under Linux up to now).  It requires flex and bison I
    think. Maybe we should ship prepared gram.c etc. like for the
    main parser too?

Jan

25 years agoPatch to allow createuser options from the command line.
Bruce Momjian [Sat, 22 Aug 1998 05:19:35 +0000 (05:19 +0000)]
Patch to allow createuser options from the command line.

25 years agoWith the attached patch, I have verified that long (> 8char anyway)
Bruce Momjian [Sat, 22 Aug 1998 04:49:05 +0000 (04:49 +0000)]
With the attached patch, I have verified that long (> 8char anyway)
usernames and passwords work correctly in both "password" and
"crypt" authorization mode.  NOTE: at least on my machine, it seems
that the crypt() routines ignore the part of the password beyond
8 characters, so there's no security gain from longer passwords in
crypt auth mode.  But they don't fail.

The login-related part of psql has apparently not been touched
since roughly the fall of Rome ;-).  It was going through huge
pushups to get around the lack of username/login parameters to
PQsetdb.  I don't know when PQsetdbLogin was added to libpq, but
it's there now ... so I was able to rip out quite a lot of crufty
code while I was at it.

It's possible that there are still bogus length limits on username
or password in some of the other PostgreSQL user interfaces besides
psql/libpq.  I will leave it to other folks to check that code.

regards, tom lane

25 years agoThe attached patch fixes a problem that I seem to have introduced
Bruce Momjian [Sat, 22 Aug 1998 04:34:22 +0000 (04:34 +0000)]
The attached patch fixes a problem that I seem to have introduced
with the new support for asynchronous NOTIFY in libpgtcl.  With
the current sources, if the backend disconnects unexpectedly then
the tcl/tk application coredumps when control next reaches the idle
loop.  Oops.

regards, tom lane

25 years agoHere are additional patches for the UnixWare 7 port.
Bruce Momjian [Sat, 22 Aug 1998 04:24:41 +0000 (04:24 +0000)]
Here are additional patches for the UnixWare 7 port.

Summary of changes:

In pqcomm.h, use the SUN_LEN macro if it is defined to calculate
the size of the sockaddr_un structure.

In unixware.h, drop the use of the UNIXWARE macro.  Everything can
be handled with the USE_UNIVEL_CC and DISABLE_COMPLEX_MACRO macros.

In s_lock.h, remove the reference to the UNIXWARE macro (see above).

In the unixware template, add the YFLAGS:-d line.

In various makefile templates, add (or cleanup) unixware and univel
port specific information.

-- Billy G. Allie

25 years agoFix display in initdb.
Bruce Momjian [Fri, 21 Aug 1998 23:22:36 +0000 (23:22 +0000)]
Fix display in initdb.

25 years agoFix for index problem that showed up in constraint test.
Bruce Momjian [Thu, 20 Aug 1998 23:01:24 +0000 (23:01 +0000)]
Fix for index problem that showed up in constraint test.

25 years agoupdate fixes.
Bruce Momjian [Thu, 20 Aug 1998 22:24:11 +0000 (22:24 +0000)]
update fixes.

25 years agofix for index problem.
Bruce Momjian [Thu, 20 Aug 1998 22:07:46 +0000 (22:07 +0000)]
fix for index problem.

25 years agoAnother vacuum fix.
Bruce Momjian [Thu, 20 Aug 1998 15:16:59 +0000 (15:16 +0000)]
Another vacuum fix.

25 years agoFix for vacuum introduced today.
Bruce Momjian [Wed, 19 Aug 1998 23:48:23 +0000 (23:48 +0000)]
Fix for vacuum introduced today.

25 years agoVacuum cleanup.
Bruce Momjian [Wed, 19 Aug 1998 22:01:18 +0000 (22:01 +0000)]
Vacuum cleanup.

25 years agoVacuum fix. Was modifying cache.
Bruce Momjian [Wed, 19 Aug 1998 19:59:49 +0000 (19:59 +0000)]
Vacuum fix.  Was modifying cache.

25 years agoFix for vacuum updating problem.
Bruce Momjian [Wed, 19 Aug 1998 15:47:36 +0000 (15:47 +0000)]
Fix for vacuum updating problem.

25 years agofix for ecpg corruption
Bruce Momjian [Wed, 19 Aug 1998 14:51:31 +0000 (14:51 +0000)]
fix for ecpg corruption

25 years agoheap_fetch requires buffer pointer, must be released; heap_getnext
Bruce Momjian [Wed, 19 Aug 1998 02:04:17 +0000 (02:04 +0000)]
heap_fetch requires buffer pointer, must be released; heap_getnext
no longer returns buffer pointer, can be gotten from scan;
descriptor; bootstrap can create multi-key indexes;
pg_procname index now is multi-key index; oidint2, oidint4, oidname
are gone (must be removed from regression tests); use System Cache
rather than sequential scan in many places; heap_modifytuple no
longer takes buffer parameter; remove unused buffer parameter in
a few other functions; oid8 is not index-able; remove some use of
single-character variable names; cleanup Buffer variables usage
and scan descriptor looping; cleaned up allocation and freeing of
tuples; 18k lines of diff;

25 years agoFix for SNPRINTF test in configure
Marc G. Fournier [Wed, 19 Aug 1998 00:16:04 +0000 (00:16 +0000)]
Fix for SNPRINTF test in configure

From: Tom Lane <tgl@sss.pgh.pa.us>

25 years agoAdd rule tests to regression tests...
Marc G. Fournier [Wed, 19 Aug 1998 00:14:39 +0000 (00:14 +0000)]
Add rule tests to regression tests...

25 years agoFrom: Jan Wieck <jwieck@debis.com>
Marc G. Fournier [Tue, 18 Aug 1998 00:49:04 +0000 (00:49 +0000)]
From: Jan Wieck <jwieck@debis.com>

Hi,

    as  proposed here comes the first patch for the query rewrite
    system.

  <for details, see archive dated Mon, 17 Aug 1998>

25 years agoMake small sentence cleanups. Add missing formatting for program example.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:22:58 +0000 (16:22 +0000)]
Make small sentence cleanups. Add missing formatting for program example.

25 years agoRename Id tags to eliminate the trailing "-1" field. This propagates
Thomas G. Lockhart [Mon, 17 Aug 1998 16:21:05 +0000 (16:21 +0000)]
Rename Id tags to eliminate the trailing "-1" field. This propagates
 into cleaner html output file names.

25 years agoMove SQL reference pages up into the User's Guide.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:20:33 +0000 (16:20 +0000)]
Move SQL reference pages up into the User's Guide.

25 years agoSmall sentence cleanups. Add tags for acronyms and products.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:18:13 +0000 (16:18 +0000)]
Small sentence cleanups. Add tags for acronyms and products.

25 years agoBring document list closer to up to day.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:17:07 +0000 (16:17 +0000)]
Bring document list closer to up to day.
Add a note on sgml-tools that they are now working with jade and so
 may become the toolset of choice in the future.

25 years agoEnhanced the allowed constraint syntax, so remove mention of the problem.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:15:31 +0000 (16:15 +0000)]
Enhanced the allowed constraint syntax, so remove mention of the problem.

25 years agoUpdate Id fields in header tags. Rename the TUTORIAL tag to avoid
Thomas G. Lockhart [Mon, 17 Aug 1998 16:14:06 +0000 (16:14 +0000)]
Update Id fields in header tags. Rename the TUTORIAL tag to avoid
 conflicting with the tutorial.sgml container document.

25 years agoFix small typo.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:12:35 +0000 (16:12 +0000)]
Fix small typo.

25 years agoUpdate the random test so it should succeed most of the time.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:11:35 +0000 (16:11 +0000)]
Update the random test so it should succeed most of the time.
Instead of directly showing the random results, test the results
 for the expected behavior (range and randomness).

25 years agoUpdate for changes to gram.y.
Thomas G. Lockhart [Mon, 17 Aug 1998 16:08:54 +0000 (16:08 +0000)]
Update for changes to gram.y.

25 years agoAllow NOT LIKE, IN, NOT IN, BETWEEN, and NOT BETWEEN expressions
Thomas G. Lockhart [Mon, 17 Aug 1998 16:08:34 +0000 (16:08 +0000)]
Allow NOT LIKE, IN, NOT IN, BETWEEN, and NOT BETWEEN expressions
 in constraint clauses.
 IN and NOT IN only allow constaints, not subselects.
Jose' Soares' new reference docs pointed out the discrepency.
 Updating the docs too...

25 years agoUse the parser macro IS_BINARY_COMPATIBLE() to allow more success
Thomas G. Lockhart [Mon, 17 Aug 1998 16:03:31 +0000 (16:03 +0000)]
Use the parser macro IS_BINARY_COMPATIBLE() to allow more success
 in type checking for DEFAULT contraint clauses.
Could do more type coersion later...

25 years agoFrom: Tom Lane <tgl@sss.pgh.pa.us>To: pgsql-patches@postgreSQL.org
Marc G. Fournier [Mon, 17 Aug 1998 03:53:37 +0000 (03:53 +0000)]
From: Tom Lane <tgl@sss.pgh.pa.us>To: pgsql-patches@postgreSQL.org

Sigh.  That tweak needs a tweak --- I didn't realize that ".DEFAULT"
processing ignores dependencies, at least in the version of gmake I
have here (not sure if it's a bug or not).  Apply this patch aftermy previous one...

25 years agoJust a couple of "after-commit" cleanups...
Marc G. Fournier [Mon, 17 Aug 1998 03:52:36 +0000 (03:52 +0000)]
Just a couple of "after-commit" cleanups...

25 years agoDate: Sun, 16 Aug 1998 14:56:48 -0400
Marc G. Fournier [Mon, 17 Aug 1998 03:50:43 +0000 (03:50 +0000)]
Date: Sun, 16 Aug 1998 14:56:48 -0400
From: Tom Lane <tgl@sss.pgh.pa.us>
Attached is a patch for this weekend's work on libpq.  I've dealt
with several issues:

        <for details: see message, in pgsql-patches archive for above data>

25 years agoFrom: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
Marc G. Fournier [Mon, 17 Aug 1998 03:35:05 +0000 (03:35 +0000)]
From: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>

Here is some more contrib-fodder, based on TIH's IP address type,
for ISBN and ISSN identifiers (which I just happened to need to keep
track of the things in my library).

25 years agoAllow a null pointer to be returned from get_opname().
Thomas G. Lockhart [Sun, 16 Aug 1998 05:38:41 +0000 (05:38 +0000)]
Allow a null pointer to be returned from get_opname().
Previously, had thrown an error, but looking for alternate strategies
 for table indices utilization would prefer to continue.

25 years agoCheck for null pointer returned from get_opname().
Thomas G. Lockhart [Sun, 16 Aug 1998 05:37:04 +0000 (05:37 +0000)]
Check for null pointer returned from get_opname().
Don't bother checking for alternate strategies if so
 since it was more likely a function or some other non-operator anyway.

25 years agoCheck for null pointer returned from get_opname().
Thomas G. Lockhart [Sun, 16 Aug 1998 05:35:35 +0000 (05:35 +0000)]
Check for null pointer returned from get_opname().

25 years agoDisable not-ready-to-use support code for the line data type.
Thomas G. Lockhart [Sun, 16 Aug 1998 04:06:55 +0000 (04:06 +0000)]
Disable not-ready-to-use support code for the line data type.
Bracket things with #ifdef ENABLE_LINE_TYPE.
The line data type has always been used internally to support other types,
 but I/O routines have never been defined for it.

25 years agoThis README only mentions nroff problems with page.5, which has been
Thomas G. Lockhart [Sat, 15 Aug 1998 16:39:21 +0000 (16:39 +0000)]
This README only mentions nroff problems with page.5, which has been
 removed from the man pages and put into the new sgml docs.

25 years agoInclude a sentence in the top pointing to the new docs.
Thomas G. Lockhart [Sat, 15 Aug 1998 16:36:22 +0000 (16:36 +0000)]
Include a sentence in the top pointing to the new docs.
pgbuiltin.3 is obsolete for sure, and libpq.3 can become so since the
 size and scope of this man page is not appropriate in a man page format.

25 years agoConverted to page.sgml. Not needed in a man page.
Thomas G. Lockhart [Sat, 15 Aug 1998 13:01:26 +0000 (13:01 +0000)]
Converted to page.sgml. Not needed in a man page.

25 years agoInformation moved to sgml source files.
Thomas G. Lockhart [Sat, 15 Aug 1998 07:03:59 +0000 (07:03 +0000)]
Information moved to sgml source files.
The "Oracle compatibility" page should have always been in with functions
 anyway. The BKI information is not really appropriate for a man page.

25 years agoMake minor formatting improvements to statement synopses.
Thomas G. Lockhart [Sat, 15 Aug 1998 07:00:37 +0000 (07:00 +0000)]
Make minor formatting improvements to statement synopses.
Make sure all files end with a newline.

25 years agoUpdate for new information on PQsetNoticeProcessor().
Thomas G. Lockhart [Sat, 15 Aug 1998 06:56:36 +0000 (06:56 +0000)]
Update for new information on PQsetNoticeProcessor().
Now back in sync with libpq.3 man page, which should be obsolete soon.

25 years agoChange Id field in chapter tag to change html output file name.
Thomas G. Lockhart [Sat, 15 Aug 1998 06:55:05 +0000 (06:55 +0000)]
Change Id field in chapter tag to change html output file name.

25 years agoInclude working list of all documentation sources, with current status
Thomas G. Lockhart [Sat, 15 Aug 1998 06:53:52 +0000 (06:53 +0000)]
Include working list of all documentation sources, with current status
 and plans for some of them.

25 years agoInclude new chapters.
Thomas G. Lockhart [Sat, 15 Aug 1998 06:52:54 +0000 (06:52 +0000)]
Include new chapters.

25 years agoNice exposition on indices and keys from Herouth Maoz which appeared
Thomas G. Lockhart [Sat, 15 Aug 1998 06:52:03 +0000 (06:52 +0000)]
Nice exposition on indices and keys from Herouth Maoz which appeared
 on the mailing lists a while ago. Maybe slightly changed to fit docs.
Will go into the User's Guide.

25 years agoConvert information from man pages. Not really suitable for man pages
Thomas G. Lockhart [Sat, 15 Aug 1998 06:49:33 +0000 (06:49 +0000)]
Convert information from man pages. Not really suitable for man pages
 anyway, so this will replace them as of now.

25 years agoRemove single-argument trim() function from table.
Thomas G. Lockhart [Sat, 15 Aug 1998 06:47:39 +0000 (06:47 +0000)]
Remove single-argument trim() function from table.
Never seen because the parser frontend converts all trim() calls to
 btrim(), ltrim(), and rtime() calls before execution.

25 years agoSome old cleanup fixes for close_ps() from G. Thaker.
Thomas G. Lockhart [Sat, 15 Aug 1998 06:45:10 +0000 (06:45 +0000)]
Some old cleanup fixes for close_ps() from G. Thaker.

25 years agoAllow binary-compatible indices to be considered when checking for valid
Thomas G. Lockhart [Fri, 14 Aug 1998 16:13:07 +0000 (16:13 +0000)]
Allow binary-compatible indices to be considered when checking for valid
 indices for restriction clauses containing a constant.
Note that if an index does not match directly (usually because the types
 on both side of the clause don't match), and if a binary-compatible index
 is identified, then the operator function will be replaced by a new
 one. Should not be a problem, but be sure that if types are listed as
 being binary compatible (in parse_coerce.h) then the comparison functions
 are also binary-compatible, giving equivalent results.

25 years agoInclude OID as a built-in type.
Thomas G. Lockhart [Fri, 14 Aug 1998 16:07:00 +0000 (16:07 +0000)]
Include OID as a built-in type.