]> granicus.if.org Git - postgresql/log
postgresql
22 years agoPatch from Ryouichi Matsuda
Dave Cramer [Tue, 5 Mar 2002 03:29:30 +0000 (03:29 +0000)]
Patch from Ryouichi Matsuda
 An attached patch corrects problem of this bug and fractional second.

 The handling of time zone was as follows:

   (a) with time zone
       using  SimpleDateFormat("yyyy-MM-dd HH:mm:ss z")
   (b) without time zone
       using  SimpleDateFormat("yyyy-MM-dd HH:mm:ss")

 About problem of fractional second,
 Fractional second was changed from milli-second to nano-second

22 years ago patch from Mitchel Friedman to fix getTables
Dave Cramer [Tue, 5 Mar 2002 03:02:52 +0000 (03:02 +0000)]
 patch from Mitchel Friedman to fix getTables

22 years agoShowing index details with \d on psql.
Bruce Momjian [Tue, 5 Mar 2002 02:42:56 +0000 (02:42 +0000)]
Showing index details with \d on psql.

Greg Sabino Mullane

22 years ago patch from Vicktor to fix Numeric decimal digits in getColumns
Dave Cramer [Tue, 5 Mar 2002 02:14:08 +0000 (02:14 +0000)]
 patch from Vicktor to fix Numeric decimal digits in getColumns

22 years agoBack out python patch:
Bruce Momjian [Tue, 5 Mar 2002 02:01:44 +0000 (02:01 +0000)]
Back out python patch:

Elliot Lee wrote:
> This patch to the python bindings adds C versions of the often-used
query
> args quoting routines, as well as support for quoting lists e.g.
> dbc.execute("SELECT * FROM foo WHERE blah IN %s", ([1,2,3],))

22 years agoThis patch to the python bindings adds C versions of the often-used
Bruce Momjian [Tue, 5 Mar 2002 00:23:13 +0000 (00:23 +0000)]
This patch to the python bindings adds C versions of the often-used
query args quoting routines, as well as support for quoting lists e.g.
dbc.execute("SELECT * FROM foo WHERE blah IN %s", ([1,2,3],))

Elliot Lee

22 years agoAdd \timing patch to psql. Times all queries.
Bruce Momjian [Tue, 5 Mar 2002 00:01:03 +0000 (00:01 +0000)]
Add \timing patch to psql.  Times all queries.

Greg Sabino Mullane

22 years agoAdd PQunescapeBytea libpq function.
Bruce Momjian [Mon, 4 Mar 2002 23:59:14 +0000 (23:59 +0000)]
Add PQunescapeBytea libpq function.

Everyone using libpq and bytea is probably having to invent this wheel..

Patrick Welche

22 years agoImprove libpgeasy API for multiple result sets, add example.
Bruce Momjian [Mon, 4 Mar 2002 18:50:21 +0000 (18:50 +0000)]
Improve libpgeasy API for multiple result sets, add example.

22 years agoAdd Assert check to catch vsnprintf overrunning its buffer. (Seen to
Tom Lane [Mon, 4 Mar 2002 18:34:02 +0000 (18:34 +0000)]
Add Assert check to catch vsnprintf overrunning its buffer.  (Seen to
occur on Solaris 7 in 64-bit mode, for one.)

22 years agoUpdate FAQ_Solaris with info about gcc 2.95.1 problems and how to work
Tom Lane [Mon, 4 Mar 2002 17:47:11 +0000 (17:47 +0000)]
Update FAQ_Solaris with info about gcc 2.95.1 problems and how to work
around 64-bit vsnprintf bug.

22 years agoMake port makefile slightly less crufty.
Tom Lane [Mon, 4 Mar 2002 17:43:32 +0000 (17:43 +0000)]
Make port makefile slightly less crufty.

22 years agoModify:
Bruce Momjian [Mon, 4 Mar 2002 16:50:36 +0000 (16:50 +0000)]
Modify:

< * Add GUC parameter for eurodates
> * Add GUC parameter for DATESTYLE

22 years agoAdd:
Bruce Momjian [Mon, 4 Mar 2002 16:20:31 +0000 (16:20 +0000)]
Add:

> * Add GUC parameter for eurodates

22 years agoFix very ancient breakage in alter-table tests: apparently, there was a
Tom Lane [Mon, 4 Mar 2002 05:17:55 +0000 (05:17 +0000)]
Fix very ancient breakage in alter-table tests: apparently, there was a
type named 'dt' back in Postgres 4.2, and the regression test wasn't
updated when it was removed.  Per report from Patricia Holben of Great
Bridge.

22 years agoIf presented db path has a trailing slash, remove it to avoid generating
Tom Lane [Mon, 4 Mar 2002 04:45:27 +0000 (04:45 +0000)]
If presented db path has a trailing slash, remove it to avoid generating
double slashes in generated filenames.  This is not strictly necessary
on standard Unixen, but I'm being a neatnik...

22 years agoFix bug in extract/date_part for milliseconds/miscroseconds and
Tatsuo Ishii [Mon, 4 Mar 2002 03:55:50 +0000 (03:55 +0000)]
Fix bug in extract/date_part for milliseconds/miscroseconds and
timestamp/timestamptz combo. Now extract/date_part returns
seconds*1000 or 1000000 + fraction part as the manual stats.
regression test are also fixed.

See the thread in pgsql-hackers:

Subject: Re: [HACKERS] timestamp_part() bug?
Date: Sat, 02 Mar 2002 11:29:53 +0900

22 years agoFix bogus error check in pg_execute, per report from lbayuk@mindspring.com.
Tom Lane [Mon, 4 Mar 2002 02:41:49 +0000 (02:41 +0000)]
Fix bogus error check in pg_execute, per report from lbayuk@mindspring.com.

22 years agoFurther work on elog cleanup: fix some bogosities in elog's logic about
Tom Lane [Mon, 4 Mar 2002 01:46:04 +0000 (01:46 +0000)]
Further work on elog cleanup: fix some bogosities in elog's logic about
when to send what to which, prevent recursion by introducing new COMMERROR
elog level for client-communication problems, get rid of direct writes
to stderr in backend/libpq files, prevent non-error elogs from going to
client during the authentication cycle.

22 years agoUpdate FAQ.
Bruce Momjian [Sun, 3 Mar 2002 18:35:46 +0000 (18:35 +0000)]
Update FAQ.

22 years agoCatcaches can now store negative entries as well as positive ones, to
Tom Lane [Sun, 3 Mar 2002 17:47:56 +0000 (17:47 +0000)]
Catcaches can now store negative entries as well as positive ones, to
speed up repetitive failed searches; per pghackers discussion in late
January.  inval.c logic substantially simplified, since we can now treat
inserts and deletes alike as far as inval events are concerned.  Some
repair work needed in heap_create_with_catalog, which turns out to have
been doing CommandCounterIncrement at a point where the new relation has
non-self-consistent catalog entries.  With the new inval code, that
resulted in assert failures during a relcache entry rebuild.

22 years agoUpdate FAQ.
Bruce Momjian [Sun, 3 Mar 2002 16:02:31 +0000 (16:02 +0000)]
Update FAQ.

22 years agoDefault server_min_messages is NOTICE.
Bruce Momjian [Sun, 3 Mar 2002 02:11:09 +0000 (02:11 +0000)]
Default server_min_messages is NOTICE.

22 years agoCause regression tests to pass again after elog changes.
Tom Lane [Sun, 3 Mar 2002 01:05:22 +0000 (01:05 +0000)]
Cause regression tests to pass again after elog changes.

22 years agoSome kibitzing about appropriate elog levels for sinval messages.
Tom Lane [Sat, 2 Mar 2002 23:35:57 +0000 (23:35 +0000)]
Some kibitzing about appropriate elog levels for sinval messages.

22 years agoCommit to match discussed elog() changes. Only update is that LOG is
Bruce Momjian [Sat, 2 Mar 2002 21:39:36 +0000 (21:39 +0000)]
Commit to match discussed elog() changes.  Only update is that LOG is
now just below FATAL in server_min_messages.  Added more text to
highlight ordering difference between it and client_min_messages.

---------------------------------------------------------------------------

REALLYFATAL => PANIC
STOP => PANIC
New INFO level the prints to client by default
New LOG level the prints to server log by default
Cause VACUUM information to print only to the client
NOTICE => INFO where purely information messages are sent
DEBUG => LOG for purely server status messages
DEBUG removed, kept as backward compatible
DEBUG5, DEBUG4, DEBUG3, DEBUG2, DEBUG1 added
DebugLvl removed in favor of new DEBUG[1-5] symbols
New server_min_messages GUC parameter with values:
        DEBUG[5-1], INFO, NOTICE, ERROR, LOG, FATAL, PANIC
New client_min_messages GUC parameter with values:
        DEBUG[5-1], LOG, INFO, NOTICE, ERROR, FATAL, PANIC
Server startup now logged with LOG instead of DEBUG
Remove debug_level GUC parameter
elog() numbers now start at 10
Add test to print error message if older elog() values are passed to elog()
Bootstrap mode now has a -d that requires an argument, like postmaster

22 years agoAdd code to allow profiling of backends on Linux: save and restore the
Tom Lane [Sat, 2 Mar 2002 20:46:12 +0000 (20:46 +0000)]
Add code to allow profiling of backends on Linux: save and restore the
profiling timer setting across fork().  The correct way to build a
profilable backend on Linux is now gmake PROFILE="-pg -DLINUX_PROFILE"

22 years agoDon't bother to request SSL connection over a Unix socket, since the
Tom Lane [Sat, 2 Mar 2002 00:49:22 +0000 (00:49 +0000)]
Don't bother to request SSL connection over a Unix socket, since the
postmaster won't accept the request anyway.  (If your kernel can't
be trusted, SSL will not help you.)

22 years agoArray slice extraction should produce a result array with index lower
Tom Lane [Sat, 2 Mar 2002 00:34:24 +0000 (00:34 +0000)]
Array slice extraction should produce a result array with index lower
bounds of 1, not the lower bound subscripts of the original slice.
Per bug report from Andre Holzner, 1-Feb-02.

22 years agoUser and database-specific session defaults for run-time configuration
Peter Eisentraut [Fri, 1 Mar 2002 22:45:19 +0000 (22:45 +0000)]
User and database-specific session defaults for run-time configuration
variables.  New commands ALTER DATABASE ... SET and ALTER USER ... SET.

22 years agoarray_ref() should set isNull to false explicitly if it's not going to
Peter Eisentraut [Fri, 1 Mar 2002 22:17:10 +0000 (22:17 +0000)]
array_ref() should set isNull to false explicitly if it's not going to
return NULL.

22 years agoFix thinko: cost_mergejoin must pay attention to which side of the
Tom Lane [Fri, 1 Mar 2002 20:50:20 +0000 (20:50 +0000)]
Fix thinko: cost_mergejoin must pay attention to which side of the
mergeclause is which when extracting selectivity info.

22 years agoRevert quotes, 7.2 doesn't need them for CREATE FUNCTION.
Bruce Momjian [Fri, 1 Mar 2002 20:38:12 +0000 (20:38 +0000)]
Revert quotes, 7.2 doesn't need them for CREATE FUNCTION.

22 years agoOops, we didn't have single quotes around 'langname' in CREATE FUNCTION
Bruce Momjian [Fri, 1 Mar 2002 19:01:22 +0000 (19:01 +0000)]
Oops, we didn't have single quotes around 'langname' in CREATE FUNCTION
manual page;  pointed out by IRC user.

22 years agoAdd files for Unicode support.
Hiroshi Inoue [Fri, 1 Mar 2002 07:18:34 +0000 (07:18 +0000)]
Add files for Unicode support.

22 years agoSecond thoughts dept: arrange to cache mergejoin scan selectivity
Tom Lane [Fri, 1 Mar 2002 06:01:20 +0000 (06:01 +0000)]
Second thoughts dept: arrange to cache mergejoin scan selectivity
in RestrictInfo nodes, instead of recomputing on every use.

22 years agoTeach planner about the idea that a mergejoin won't necessarily read
Tom Lane [Fri, 1 Mar 2002 04:09:28 +0000 (04:09 +0000)]
Teach planner about the idea that a mergejoin won't necessarily read
both input streams to the end.  If one variable's range is much less
than the other, an indexscan-based merge can win by not scanning all
of the other table.  Per example from Reinhard Max.

22 years agoTweak pg_exec_query_string so that we close down transaction command
Tom Lane [Wed, 27 Feb 2002 23:16:07 +0000 (23:16 +0000)]
Tweak pg_exec_query_string so that we close down transaction command
before reporting command-complete message for the final command of a
query string.  This way, any errors detected during finish_xact_command
(such as RI violations) will appear to be part of the final command,
rather than coming out after the command is reported complete.  This
avoids confusing PQendcopy and other not-overly-bright clients.
Per Lee Harr's bug report of 25-Feb-02.

22 years agoSome cleanups in CREATE/ALTER/DROP USER ref pages.
Peter Eisentraut [Wed, 27 Feb 2002 21:14:54 +0000 (21:14 +0000)]
Some cleanups in CREATE/ALTER/DROP USER ref pages.

22 years ago#ifdef out pg_dump's check on whether a sequence's sequence_name field
Tom Lane [Wed, 27 Feb 2002 20:59:05 +0000 (20:59 +0000)]
#ifdef out pg_dump's check on whether a sequence's sequence_name field
matches the sequence name from pg_class.  This fails if the sequence has
been renamed, and seems rather pointless in any case.
Also improve a couple of error messages about inconsistencies.

22 years agoParanoia about data structure lifetime ...
Tom Lane [Wed, 27 Feb 2002 19:52:41 +0000 (19:52 +0000)]
Paranoia about data structure lifetime ...

22 years agoClean up BeginCommand and related routines. BeginCommand and EndCommand
Tom Lane [Wed, 27 Feb 2002 19:36:13 +0000 (19:36 +0000)]
Clean up BeginCommand and related routines.  BeginCommand and EndCommand
are now both invoked once per received SQL command (raw parsetree) from
pg_exec_query_string.  BeginCommand is actually just an empty routine
at the moment --- all its former operations have been pushed into tuple
receiver setup routines in printtup.c.  This makes for a clean distinction
between BeginCommand/EndCommand (once per command) and the tuple receiver
setup/teardown routines (once per ExecutorRun call), whereas the old code
was quite ad hoc.  Along the way, clean up the calling conventions for
ExecutorRun a little bit.

22 years agoUpdate FAQ.
Bruce Momjian [Wed, 27 Feb 2002 04:52:19 +0000 (04:52 +0000)]
Update FAQ.

22 years agoImprove the handling of ODBC escape(a request from Marcelo Aceto).
Hiroshi Inoue [Wed, 27 Feb 2002 01:43:24 +0000 (01:43 +0000)]
Improve the handling of ODBC escape(a request from Marcelo Aceto).
Change SQLGetFunctions() to reply not yet implemented ODBC 3.0
functions precisely.

22 years agoFix bug in COPY FROM when DELIMITER is not in ASCII range.
Tatsuo Ishii [Wed, 27 Feb 2002 01:34:41 +0000 (01:34 +0000)]
Fix bug in COPY FROM when DELIMITER is not in ASCII range.
See pgsql-bugs/pgsql-hackers discussion "COPY FROM is not 8bit clean"
around 2002/02/26 for more details -- Tatsuo Ishii

22 years agoRestructure command-completion-report code so that there is just one
Tom Lane [Tue, 26 Feb 2002 22:47:12 +0000 (22:47 +0000)]
Restructure command-completion-report code so that there is just one
report for each received SQL command, regardless of rewriting activity.
Also ensure that this report comes from the 'original' command, not the
last command generated by rewrite; this fixes 7.2 breakage for INSERT
commands that have actions added by rules.  Fernando Nasser and Tom Lane.

22 years agoCorrect typo noticed by Christopher Kings-Lynne.
Tom Lane [Tue, 26 Feb 2002 15:59:08 +0000 (15:59 +0000)]
Correct typo noticed by Christopher Kings-Lynne.

22 years agoImplementation for cancelQuery by Grant Finnemore <grantf@guruhut.co.za>
Dave Cramer [Tue, 26 Feb 2002 02:15:55 +0000 (02:15 +0000)]
Implementation for cancelQuery by Grant Finnemore <grantf@guruhut.co.za>

22 years agoUse quote-marks to try to clarify a plpgsql error message, per
Tom Lane [Tue, 26 Feb 2002 00:00:08 +0000 (00:00 +0000)]
Use quote-marks to try to clarify a plpgsql error message, per
gripe from Dan Langille.

22 years agoTweak psql's \connect command to not downcase unquoted database and user
Tom Lane [Mon, 25 Feb 2002 21:37:42 +0000 (21:37 +0000)]
Tweak psql's \connect command to not downcase unquoted database and user
names.  This is a temporary measure to allow backwards compatibility with
7.2 and earlier pg_dump.  7.2.1 and later pg_dump will double-quote mixed
case names in \connect.  Once we feel that older dumps are not a problem
anymore, we can revert this change and treat \connect arguments as normal
SQL identifiers.

22 years agoAdded:
Bruce Momjian [Mon, 25 Feb 2002 20:30:19 +0000 (20:30 +0000)]
Added:

> * Remove memory/file descriptor freeing befor elog(ERROR)  (Bruce)

22 years agoUpdate FAQ_DEV.
Bruce Momjian [Mon, 25 Feb 2002 20:29:39 +0000 (20:29 +0000)]
Update FAQ_DEV.

22 years agoUpdate FAQ.
Bruce Momjian [Mon, 25 Feb 2002 20:29:22 +0000 (20:29 +0000)]
Update FAQ.

22 years agoFix for PAM error message display:
Bruce Momjian [Mon, 25 Feb 2002 20:07:02 +0000 (20:07 +0000)]
Fix for PAM error message display:

> and that the right fix is to make each of the subsequent calls be in
> this same pattern, not to try to emulate their nonsensical style.

Dominic J. Eidson

22 years agoAdd items and JDBC maintainers:
Bruce Momjian [Mon, 25 Feb 2002 18:48:54 +0000 (18:48 +0000)]
Add items and JDBC maintainers:

>  o Add LISTEN/NOTIFY support to the JDBC driver (Barry)
479a481
> * Barry is Barry Lind <barry@xythos.com>
482a485
> * Dave is Dave Cramer <dave@fastcrypt.com>

22 years agoAdd note that BLCKSZ must be a power of 2.
Tom Lane [Mon, 25 Feb 2002 17:11:22 +0000 (17:11 +0000)]
Add note that BLCKSZ must be a power of 2.

22 years agoAdd a large number of time zones to the lookup table.
Thomas G. Lockhart [Mon, 25 Feb 2002 16:17:04 +0000 (16:17 +0000)]
Add a large number of time zones to the lookup table.
Fix a few apparently-wrong TZ vs DTZ declarations.

22 years agoUpdate completed items:
Bruce Momjian [Mon, 25 Feb 2002 06:23:54 +0000 (06:23 +0000)]
Update completed items:

> * -Prevent SIGHUP and 'pg_ctl reload' from changing command line
> * -Remove LIMIT #,# and force use LIMIT and OFFSET clauses in 7.3 (Bruce)
> * -Prevent create/drop scripts from allowing extra args (Bruce)
> * -Have pg_dump -C dump database location and encoding information

22 years agoThis patch will allow arbitrary levels of analyze / rewriting
Bruce Momjian [Mon, 25 Feb 2002 04:21:55 +0000 (04:21 +0000)]
This patch will allow arbitrary levels of analyze / rewriting
by making the static variables extra_before and extra_after
automatic so we can use recursion.

It gets much easier to generate extra commands now, and one can rest
assured that the extra commands will be properly analyzed/rewritten.

Without this patch, if a command produced by transformation tries to
use these static lists their first contents would be lost with
unpredictable results.  I know I could fix this by just using nconc()
instead of assignments, but the resulting order of the commands would
not be exactly what one could expect.

--
Fernando Nasser

22 years agoTry to make makefile actually work ...
Tom Lane [Mon, 25 Feb 2002 04:16:58 +0000 (04:16 +0000)]
Try to make makefile actually work ...

22 years agoFix typo in Makefile.
Bruce Momjian [Mon, 25 Feb 2002 04:08:03 +0000 (04:08 +0000)]
Fix typo in Makefile.

22 years agoI've attached a simple patch which should improve the performance of
Bruce Momjian [Mon, 25 Feb 2002 04:06:52 +0000 (04:06 +0000)]
I've attached a simple patch which should improve the performance of
hashname() and reduce the penalty incured when NAMEDATALEN is increased.
I posted this to -hackers a couple days ago, and there haven't been any
major complaints. It passes the regression tests. See -hackers for more
discussion, as well as the suggestion from Tom Lane on which this patch
is based.

Unless anyone sees any problems, please apply for 7.3.

Cheers,

Neil Conway

22 years agoFix typo in filename.
Bruce Momjian [Mon, 25 Feb 2002 03:59:39 +0000 (03:59 +0000)]
Fix typo in filename.

22 years agoAdd integer aggregator to /contrib.
Bruce Momjian [Mon, 25 Feb 2002 03:45:27 +0000 (03:45 +0000)]
Add integer aggregator to /contrib.

mlw

22 years ago*** empty log message ***
Bruce Momjian [Mon, 25 Feb 2002 03:37:36 +0000 (03:37 +0000)]
*** empty log message ***

22 years agoRe-add equals documentation with CREATE DATABASE.
Bruce Momjian [Mon, 25 Feb 2002 03:37:14 +0000 (03:37 +0000)]
Re-add equals documentation with CREATE DATABASE.

22 years agoRemove documentation of equals in CREATE DATABASE. Mention removal of
Bruce Momjian [Mon, 25 Feb 2002 02:53:46 +0000 (02:53 +0000)]
Remove documentation of equals in CREATE DATABASE.  Mention removal of
equals hack for the future.

22 years agoFix bogus createdb option syntax recommended by initlocation message.
Tom Lane [Sun, 24 Feb 2002 23:59:36 +0000 (23:59 +0000)]
Fix bogus createdb option syntax recommended by initlocation message.

22 years agoFix copy-and-paste typo.
Tom Lane [Sun, 24 Feb 2002 23:55:31 +0000 (23:55 +0000)]
Fix copy-and-paste typo.

22 years agoAdd -O/--owner switch to createdb script, in support of new OWNER option
Tom Lane [Sun, 24 Feb 2002 23:27:58 +0000 (23:27 +0000)]
Add -O/--owner switch to createdb script, in support of new OWNER option
for CREATE DATABASE.

22 years agoModify pg_dumpall so that output script uses new OWNER option of CREATE
Tom Lane [Sun, 24 Feb 2002 21:57:23 +0000 (21:57 +0000)]
Modify pg_dumpall so that output script uses new OWNER option of CREATE
DATABASE; also make it use SET SESSION AUTHORIZATION commands rather than
\connect commands.  This makes it possible to restore databases belonging
to users who do not have CREATEDB privilege.  It should also become at
least somewhat feasible to run the restore script under password
authentication --- you'll get one superuser password prompt per database,
rather than a large number of challenges for passwords belonging to
varying unspecified user names.

22 years agoAdd OWNER option to CREATE DATABASE, so superusers can create databases
Tom Lane [Sun, 24 Feb 2002 20:20:21 +0000 (20:20 +0000)]
Add OWNER option to CREATE DATABASE, so superusers can create databases
on behalf of unprivileged users.  Also, make '=' optional in CREATE
DATABASE syntax.  From Gavin Sherry, with kibitzing and docs by Tom Lane.

22 years agocompiles correctly but still doesn't work with jdk 1.4
Dave Cramer [Sun, 24 Feb 2002 16:12:51 +0000 (16:12 +0000)]
compiles correctly but still doesn't work with jdk 1.4

22 years agopgindent copy.c. Patch wasn't in proper format.
Bruce Momjian [Sun, 24 Feb 2002 02:33:33 +0000 (02:33 +0000)]
pgindent copy.c.  Patch wasn't in proper format.

22 years agoFree files on dir open failure in COPY.
Bruce Momjian [Sun, 24 Feb 2002 02:32:26 +0000 (02:32 +0000)]
Free files on dir open failure in COPY.

22 years agoAdd -N option which disables updation of branches and tellers tables.
Tatsuo Ishii [Sun, 24 Feb 2002 00:17:57 +0000 (00:17 +0000)]
Add -N option which disables updation of branches and tellers tables.
So there would be less contention with -N, that might be more realistic
test.

22 years agoAdded:
Bruce Momjian [Sat, 23 Feb 2002 21:53:37 +0000 (21:53 +0000)]
Added:

> * Add checks for missing parameters to shell script, to prevent
>   over-shifting

22 years agoOK'ed by Peter.
Bruce Momjian [Sat, 23 Feb 2002 21:49:10 +0000 (21:49 +0000)]
OK'ed by Peter.

The attached patch enables plperl to build under Cygwin.  It is
basically yet another BE_DLLLIBS patch with a perl MakeMaker twist.  I
tried the patch under Red Hat 7.1 Linux too and I did not observe any
ill effects.

Jason Tishler

22 years agoAdd fstat / S_ISDIR checks to make sure we're not trying to use a
Bruce Momjian [Sat, 23 Feb 2002 21:46:03 +0000 (21:46 +0000)]
Add fstat / S_ISDIR checks to make sure we're not trying to use a
directory for COPY TO/FROM.

Brent Verner

22 years agoFeature patch per prev thread for --with-tcl, only affects AIX
Bruce Momjian [Sat, 23 Feb 2002 21:25:35 +0000 (21:25 +0000)]
Feature patch per prev thread for --with-tcl, only affects AIX

patch is low risc, thus could be applied now, but can also wait for 7.3
Old Makefile shows, that -bnoentry is available since 4.1 .

Andreas Zeugswetter

22 years agoGuard against createdb --location=PGDATA foo; without this, the code
Tom Lane [Sat, 23 Feb 2002 20:55:46 +0000 (20:55 +0000)]
Guard against createdb --location=PGDATA foo; without this, the code
tries to create a symlink pointing at itself.  Per trouble report from
Kenneth McDowell.

22 years agoUpdate FAQ_DEV.
Bruce Momjian [Sat, 23 Feb 2002 20:09:40 +0000 (20:09 +0000)]
Update FAQ_DEV.

22 years agoAdd type conversion discussion.
Bruce Momjian [Sat, 23 Feb 2002 05:06:04 +0000 (05:06 +0000)]
Add type conversion discussion.

22 years agoHeimdal support (Kerberos V implementation from KTH)
Peter Eisentraut [Sat, 23 Feb 2002 04:17:47 +0000 (04:17 +0000)]
Heimdal support (Kerberos V implementation from KTH)

22 years agoAdd -Wno-error to CFLAGS, so the rest of the tree can compile with
Peter Eisentraut [Sat, 23 Feb 2002 04:16:04 +0000 (04:16 +0000)]
Add -Wno-error to CFLAGS, so the rest of the tree can compile with
-Werror.

22 years agoSave source of GUC settings, allowing different sources to be processed in
Peter Eisentraut [Sat, 23 Feb 2002 01:31:37 +0000 (01:31 +0000)]
Save source of GUC settings, allowing different sources to be processed in
any order without affecting results.

22 years agoMake factorial(0) return 1, as per spec.
Bruce Momjian [Sat, 23 Feb 2002 01:01:30 +0000 (01:01 +0000)]
Make factorial(0) return 1, as per spec.

22 years agoI've written on like that a while ago:
Bruce Momjian [Fri, 22 Feb 2002 23:05:35 +0000 (23:05 +0000)]
I've written on like that a while ago:

http://webmail.postgresql.org/~petere/dbsize.html

The tarball can be rolled into contrib -- now that I think of it I don't
know why I never did that.

Never imagined this would have anything to do with that TODO item,
though.
I figured oid2name accomplished that.

Peter Eisentraut   peter_e@gmx.net

22 years agoRecord for future use if we ever change the protocol:
Bruce Momjian [Fri, 22 Feb 2002 16:21:21 +0000 (16:21 +0000)]
Record for future use if we ever change the protocol:

> * Allow psql to show transaction status if backend protocol changes made

22 years agoWe had a problem with to compile pgsql-7.2 under SW-8.0.
Bruce Momjian [Fri, 22 Feb 2002 15:39:42 +0000 (15:39 +0000)]
We had a problem with to compile pgsql-7.2 under SW-8.0.
In the mailing lists I found no informations.
See note for further informations.

Add missing AuthBlockSig.

regards Heiko

22 years agoAdd Russian FAQ from Viktor Vislobokov.
Bruce Momjian [Fri, 22 Feb 2002 12:48:55 +0000 (12:48 +0000)]
Add Russian FAQ from Viktor Vislobokov.

22 years agoPlease, apply attached patch of contrib/btree_gist to 7.2.1 and current
Bruce Momjian [Fri, 22 Feb 2002 05:47:50 +0000 (05:47 +0000)]
Please, apply attached patch of contrib/btree_gist to 7.2.1 and current
cvs.
The patch fixes memory leak during creation GiST index on timestamp
column.

Thank you.

--
Teodor Sigaev
teodor@stack.net

22 years agoModify:
Bruce Momjian [Fri, 22 Feb 2002 05:42:44 +0000 (05:42 +0000)]
Modify:

> * Inline simple SQL functions to avoid overhead (Tom)
> * Precompile SQL functions to avoid overhead (Neil Conway)

22 years agoAdded:
Bruce Momjian [Fri, 22 Feb 2002 05:22:50 +0000 (05:22 +0000)]
Added:

< * Inline simple SQL functions to avoid overhead (Tom)
> * Inline simple SQL functions to avoid overhead (Neil Conway)

22 years agoAdd:
Bruce Momjian [Fri, 22 Feb 2002 04:19:37 +0000 (04:19 +0000)]
Add:

> * Inline simple SQL functions to avoid overhead (Tom)

22 years agofixed compile error
Dave Cramer [Fri, 22 Feb 2002 03:22:31 +0000 (03:22 +0000)]
fixed compile error

22 years agoAdd NAMEDATALEN discussion.
Bruce Momjian [Fri, 22 Feb 2002 03:05:26 +0000 (03:05 +0000)]
Add NAMEDATALEN discussion.

<   first, breaks notify API;
>   first, breaks notify API;  [namedatalen]

22 years agoAdd discussion.
Bruce Momjian [Fri, 22 Feb 2002 03:04:48 +0000 (03:04 +0000)]
Add discussion.

22 years agoThis patch fixes the exception thrown to inform the user the method
Dave Cramer [Fri, 22 Feb 2002 02:58:24 +0000 (02:58 +0000)]
This patch fixes the exception thrown to inform the user the method
getColumnClassName(int) is not implemented. This will futher fixes method
ResultSet.getObject(int) since it requires the getColumnClassName(int) method to return the proper java class used to map the database column.

auther Ed Yu

22 years agoPatch from Cormac Twomey
Dave Cramer [Fri, 22 Feb 2002 02:40:09 +0000 (02:40 +0000)]
Patch from Cormac Twomey
fixes getIndexInfo throwing NullPointerException
fixes getIndexInfo improper results when multiple key indexs are used