]> granicus.if.org Git - postgresql/log
postgresql
23 years agoAdd index. Only some parts of the manual set have index entries so far...
Peter Eisentraut [Sat, 12 May 2001 22:51:36 +0000 (22:51 +0000)]
Add index.  Only some parts of the manual set have index entries so far...

23 years agoModify RelationGetBufferForTuple() so that we only do lseek and lock
Tom Lane [Sat, 12 May 2001 19:58:28 +0000 (19:58 +0000)]
Modify RelationGetBufferForTuple() so that we only do lseek and lock
when we need to move to a new page; as long as we can insert the new
tuple on the same page as before, we only need LockBuffer and not the
expensive stuff.  Also, twiddle bufmgr interfaces to avoid redundant
lseeks in RelationGetBufferForTuple and BufferAlloc.  Successive inserts
now require one lseek per page added, rather than one per tuple with
several additional ones at each page boundary as happened before.
Lock contention when multiple backends are inserting in same table
is also greatly reduced.

23 years agoAdd provisions for using strdup replacement in the places that still
Peter Eisentraut [Sat, 12 May 2001 19:49:48 +0000 (19:49 +0000)]
Add provisions for using strdup replacement in the places that still
needed it.

from our fearless Ultrix porter, Alexander Klimov <ask@wisdom.weizmann.ac.il>

23 years agoAdd command '\pset footer' to psql to turn off default "(x rows)" footer.
Peter Eisentraut [Sat, 12 May 2001 19:44:46 +0000 (19:44 +0000)]
Add command '\pset footer' to psql to turn off default "(x rows)" footer.

23 years agoPL/Python should build portably now, if you can get over the fact that
Peter Eisentraut [Sat, 12 May 2001 17:49:32 +0000 (17:49 +0000)]
PL/Python should build portably now, if you can get over the fact that
there's no shared libpython.  Test suite works as well. Also, add some
documentation.

23 years agoAdd comment to explain need for bizarre-looking coding in HandleSlashCmds.
Tom Lane [Sat, 12 May 2001 17:37:15 +0000 (17:37 +0000)]
Add comment to explain need for bizarre-looking coding in HandleSlashCmds.

23 years agoMake bootstrap debug messages more readable. Clean up some clutter.
Peter Eisentraut [Sat, 12 May 2001 01:48:49 +0000 (01:48 +0000)]
Make bootstrap debug messages more readable.  Clean up some clutter.

23 years agoPL/Python integration: support in create/droplang, add CVS keywords,
Peter Eisentraut [Sat, 12 May 2001 01:30:30 +0000 (01:30 +0000)]
PL/Python integration:  support in create/droplang, add CVS keywords,
remove useless files, beat some sense into Makefile.  For me it builds
and sort of runs, so it's a start.

23 years ago - Don't dump COMMENTs in data-only dumps
Philip Warner [Sat, 12 May 2001 01:03:59 +0000 (01:03 +0000)]
 - Don't dump COMMENTs in data-only dumps
 - Fix view dumping SQL for V7.0
 - Fix bug when getting view oid with long view names
 - Treat SEQUENCE SET TOC entries as data entries rather than schema
   entries.
 - Make allowance for data entries that did not have a data dumper
   routine (eg. SEQUENCE SET)

23 years agoCleanups of pltcl unknown thingy.
Peter Eisentraut [Fri, 11 May 2001 23:38:06 +0000 (23:38 +0000)]
Cleanups of pltcl unknown thingy.

23 years agono more Andy Piper tools
Peter Eisentraut [Fri, 11 May 2001 22:27:12 +0000 (22:27 +0000)]
no more Andy Piper tools

23 years agoImprove discussion of %TYPE and %ROWTYPE.
Tom Lane [Fri, 11 May 2001 06:10:44 +0000 (06:10 +0000)]
Improve discussion of %TYPE and %ROWTYPE.

23 years agoSuppress failures in parallel regress tests due to use of same table
Tom Lane [Fri, 11 May 2001 05:09:03 +0000 (05:09 +0000)]
Suppress failures in parallel regress tests due to use of same table
name in two different tests.  This solution does not meet with universal
approval, so it may get changed later ...

23 years agoStamp CVS as 7.2. Update all interface version numbers. This is the
Bruce Momjian [Fri, 11 May 2001 01:46:33 +0000 (01:46 +0000)]
Stamp CVS as 7.2.  Update all interface version numbers.  This is the
time to do it, not during beta because people are using this stuff in
production sometimes.

23 years agoStamp for 7.1.2. Ready when you are...
Bruce Momjian [Fri, 11 May 2001 01:34:14 +0000 (01:34 +0000)]
Stamp for 7.1.2.  Ready when you are...

23 years agoUpdate FAQ.
Bruce Momjian [Fri, 11 May 2001 01:33:01 +0000 (01:33 +0000)]
Update FAQ.

23 years agoBSD/OS has new precision, and negative zeros now.
Bruce Momjian [Thu, 10 May 2001 23:46:25 +0000 (23:46 +0000)]
BSD/OS has new precision, and negative zeros now.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 23:35:50 +0000 (23:35 +0000)]
Update TODO list.

23 years agoUp version to force initdb to fix regression problems.
Bruce Momjian [Thu, 10 May 2001 22:39:24 +0000 (22:39 +0000)]
Up version to force initdb to fix regression problems.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 22:36:52 +0000 (22:36 +0000)]
Update TODO list.

23 years agoAvoid unnecessary lseek() calls by cleanups in md.c. mdfd_lstbcnt was
Tom Lane [Thu, 10 May 2001 20:38:49 +0000 (20:38 +0000)]
Avoid unnecessary lseek() calls by cleanups in md.c.  mdfd_lstbcnt was
not being consulted anywhere, so remove it and remove the _mdnblocks()
calls that were used to set it.  Change smgrextend interface to pass in
the target block number (ie, current file length) --- the caller always
knows this already, having already done smgrnblocks(), so it's silly to
do it over again inside mdextend.  Net result: extension of a file now
takes one lseek(SEEK_END) and a write(), not three lseeks and a write.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 20:28:48 +0000 (20:28 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 20:26:45 +0000 (20:26 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 17:29:55 +0000 (17:29 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Thu, 10 May 2001 17:20:10 +0000 (17:20 +0000)]
Update TODO list.

23 years agoAdd mention in README of Oracle.
Bruce Momjian [Thu, 10 May 2001 15:52:27 +0000 (15:52 +0000)]
Add mention in README of Oracle.

23 years agoAdd oracle conversion utility ora2pg in /contrib/oracle.
Bruce Momjian [Thu, 10 May 2001 15:51:05 +0000 (15:51 +0000)]
Add oracle conversion utility ora2pg in /contrib/oracle.

23 years agoAdd dbase conversion utility to /contrib.
Bruce Momjian [Thu, 10 May 2001 14:41:23 +0000 (14:41 +0000)]
Add dbase conversion utility to /contrib.

23 years agoMove disabled message up to a higher makefile.
Bruce Momjian [Thu, 10 May 2001 03:00:11 +0000 (03:00 +0000)]
Move disabled message up to a higher makefile.

23 years agoAnother add.
Bruce Momjian [Wed, 9 May 2001 23:17:23 +0000 (23:17 +0000)]
Another add.

23 years agoAdd new emails.
Bruce Momjian [Wed, 9 May 2001 23:15:55 +0000 (23:15 +0000)]
Add new emails.

23 years agoFirst cut at making indexscan cost estimates depend on correlation
Tom Lane [Wed, 9 May 2001 23:13:37 +0000 (23:13 +0000)]
First cut at making indexscan cost estimates depend on correlation
between index order and table order.

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 9 May 2001 23:11:54 +0000 (23:11 +0000)]
Update TODO list.

23 years agoThis directory contains a module that implements the "Metaphone" code as
Bruce Momjian [Wed, 9 May 2001 23:00:44 +0000 (23:00 +0000)]
This directory contains a module that implements the "Metaphone" code as
a PostgreSQL user-defined function.  The Metaphone system is a method of
matching similar sounding names (or any words) to the same code.

Metaphone was invented by Lawrence Philips as an improvement to the popular
name-hashing routine, Soundex.

This metaphone code is from Michael Kuhn, and is detailed at
   http://aspell.sourceforge.net/metaphone/metaphone-kuhn.txt

Joel Burton

23 years agoAdd --echo option to createlang and droplang.
Peter Eisentraut [Wed, 9 May 2001 22:08:19 +0000 (22:08 +0000)]
Add --echo option to createlang and droplang.
from Oliver Elphick

23 years agoWow, I am getting good at this Makefile stuff. Peter may be OK with it.
Bruce Momjian [Wed, 9 May 2001 21:42:29 +0000 (21:42 +0000)]
Wow, I am getting good at this Makefile stuff.  Peter may be OK with it.

23 years agoI think I got the pltcl Makefile unknown rules working properly.
Bruce Momjian [Wed, 9 May 2001 21:35:11 +0000 (21:35 +0000)]
I think I got the pltcl Makefile unknown rules working properly.

23 years agoI have modifed heap.c so that it won't automatically generate duplicate
Bruce Momjian [Wed, 9 May 2001 21:13:35 +0000 (21:13 +0000)]
I have modifed heap.c so that it won't automatically generate duplicate
constraint names.

> > A reasonable interpretation of DROP CONSTRAINT "foo" is to drop *all*
> > constraints named "foo" on the target table.
>
> Then it should probably be a good thing to avoid the automatic
> generation of
> duplicate names?  I might take a look at that, actually...
>

Christopher Kings-Lynne

23 years agothat's just me again, here's normal patch for KOI8_U to
Bruce Momjian [Wed, 9 May 2001 21:11:26 +0000 (21:11 +0000)]
that's just me again, here's normal patch for KOI8_U to
jdbc/Connection.java

Andy
P.S. in Connection.java if encoding=="WIN" then dbEncoding is set to
"Cp1252".
What if it's Cyrillic "WIN"? Than it should be "Cp1251". Is there any
way to fix that without making different "WIN" encodings in
PostgreSQL?

Andy Rysin

23 years ago This patch should catch cases where the types
Bruce Momjian [Wed, 9 May 2001 21:10:39 +0000 (21:10 +0000)]
    This patch should catch cases where the types
in referencing and referenced columns of an fk constraint
aren't comparable using '=' at constraint definition time
rather than insert/update time.

Stephan Szabo

23 years agoUpdate TODO list.
Bruce Momjian [Wed, 9 May 2001 20:40:30 +0000 (20:40 +0000)]
Update TODO list.

23 years agoMore cleanups to generate tcl unknown from Makefile.
Bruce Momjian [Wed, 9 May 2001 20:30:09 +0000 (20:30 +0000)]
More cleanups to generate tcl unknown from Makefile.

23 years agoautoconf
Bruce Momjian [Wed, 9 May 2001 20:20:09 +0000 (20:20 +0000)]
autoconf

23 years agoMake tcl unknown configure from Makefile, not configure.
Bruce Momjian [Wed, 9 May 2001 20:19:30 +0000 (20:19 +0000)]
Make tcl unknown configure from Makefile, not configure.

23 years agoMore --enable-pltcl-unknown renames.
Bruce Momjian [Wed, 9 May 2001 20:08:08 +0000 (20:08 +0000)]
More --enable-pltcl-unknown renames.

23 years agoChange to enable_pltcl_unknown mention.
Bruce Momjian [Wed, 9 May 2001 19:57:33 +0000 (19:57 +0000)]
Change to enable_pltcl_unknown mention.

23 years agoAdd TODO file.
Bruce Momjian [Wed, 9 May 2001 19:55:22 +0000 (19:55 +0000)]
Add TODO file.

23 years agoAdd plpython code.
Bruce Momjian [Wed, 9 May 2001 19:54:38 +0000 (19:54 +0000)]
Add plpython code.

23 years agoautoconf
Bruce Momjian [Wed, 9 May 2001 19:42:26 +0000 (19:42 +0000)]
autoconf

23 years agoChange --with-pltcl-unknown to --enable-pltcl-unknown
Bruce Momjian [Wed, 9 May 2001 19:41:57 +0000 (19:41 +0000)]
Change --with-pltcl-unknown to  --enable-pltcl-unknown

23 years agoNoticed a small bug in the code. Probably been there for some time. Note
Bruce Momjian [Wed, 9 May 2001 19:28:31 +0000 (19:28 +0000)]
Noticed a small bug in the code. Probably been there for some time. Note
that the original code would consider things like UNIX domain sockets are
regular files.

Gavin Sherry

23 years agoautoconf
Bruce Momjian [Wed, 9 May 2001 19:21:14 +0000 (19:21 +0000)]
autoconf

23 years agoThis patch adds a new configure option --with-pltcl-unknown which
Bruce Momjian [Wed, 9 May 2001 19:19:00 +0000 (19:19 +0000)]
This patch adds a new configure option --with-pltcl-unknown which
enables pltcl unknown support.

Also it adds substituting of tclsh with tclsh that was by configure in
pltcl_*mod scripts. For example, On freebsd, tclsh can be called
tclsh8.2 or
tclsh8.3 depending on installed version of Tcl.

After patching files
  src/pl/tcl/modules/pltcl_listmod
  src/pl/tcl/modules/pltcl_loadmod
  src/pl/tcl/modules/pltcl_delmod
must be renamed(copied,repocopied) to
  src/pl/tcl/modules/pltcl_listmod.in
  src/pl/tcl/modules/pltcl_loadmod.in
  src/pl/tcl/modules/pltcl_delmod.in

seva@sevasoft.kiev.ua

23 years agoRight-align \du user-id.
Bruce Momjian [Wed, 9 May 2001 17:57:42 +0000 (17:57 +0000)]
Right-align \du user-id.

23 years agoFix memory leak in new psql \du code.
Bruce Momjian [Wed, 9 May 2001 17:49:42 +0000 (17:49 +0000)]
Fix memory leak in new psql \du code.

23 years agoFix libpq++'s FieldSize to return int, not short.
Bruce Momjian [Wed, 9 May 2001 17:46:11 +0000 (17:46 +0000)]
Fix libpq++'s FieldSize to return int, not short.

23 years agoAdd mention of getLength returning short.
Bruce Momjian [Wed, 9 May 2001 17:37:17 +0000 (17:37 +0000)]
Add mention of getLength returning short.

23 years agoHere's a version of my suggested diffs transplanted to 7.1 beta 5. I'm
Bruce Momjian [Wed, 9 May 2001 17:29:10 +0000 (17:29 +0000)]
Here's a version of my suggested diffs transplanted to 7.1 beta 5.  I'm
still looking at the best way to integrate Tom Vijlbrief's fixes
(insofar as they're still needed); would 7.2 be a suitable time for
incompatible API changes?

Jeroen

Changes:

(*) Introduced bool, true, false (replacing some int, 1, 0)
(*) Made some member functions const
(*) Documented GetIsNull()
(*) Marked DisplayTuples() and PrintTuples() as obsolescent; fixed possible
    portability problem (assumed that NULL pointer equals all-zero bit pattern)
(*) PrintTuples(): renamed width parameter to fillAlign to conform with other
    usage; fixed memory leak and compile issue w.r.t. field separator (should
    also slightly improve performance)
(*) Fixed some minor compilation issues
(*) Moved "using namespace std;" out of headers, where they didn't belong; used
    new (temporary) preprocessor macro PGSTD to do this
(*) Made ToString() static, removed unneeded memset(), made buffer size adapt
    to sizeof(int)
(*) Made some constructors explicit
(*) Changed some const std::string & parameters to plain std::string
(*) Marked PgCursor::Cursor(std::string) as obsolescent (setter with same name
    as getter--bad style)
(*) Renamed some paramaters previously named "string"
(*) Introduced size_type typedef for number of tuples in result set
(*) PgTransaction now supports re-opening after closing, and aborts if not
    explicitly committed prior to destruction

J. T. Vermeulen

23 years agoUse ColId instead of Ident for SET SESSION AUTHORIZATION.
Peter Eisentraut [Wed, 9 May 2001 16:50:44 +0000 (16:50 +0000)]
Use ColId instead of Ident for SET SESSION AUTHORIZATION.

23 years agoMention new jdbc mailing list instead of interfaces list.
Bruce Momjian [Wed, 9 May 2001 15:51:37 +0000 (15:51 +0000)]
Mention new jdbc mailing list instead of interfaces list.

23 years agoIf it's true that the ALTER TABLE x ADD CONSTRAINT x CHECK (x) syntax is
Bruce Momjian [Wed, 9 May 2001 13:27:15 +0000 (13:27 +0000)]
If it's true that the ALTER TABLE x ADD CONSTRAINT x CHECK (x) syntax is
supported in 7.1.1, here is a patch to that alter_table.sgml that documents
it.

Christopher Kings-Lynne

23 years agoCause planner to make use of average-column-width statistic that is now
Tom Lane [Wed, 9 May 2001 00:35:09 +0000 (00:35 +0000)]
Cause planner to make use of average-column-width statistic that is now
collected by ANALYZE.  Also, add some modest amount of intelligence to
guesses that are used for varlena columns in the absence of any ANALYZE
statistics.  The 'width' reported by EXPLAIN is finally something less
than totally bogus for varlena columns ... and, in consequence, hashjoin
estimating should be a little better ...

23 years agoUpdate runtime shared memory computations.
Bruce Momjian [Tue, 8 May 2001 21:09:47 +0000 (21:09 +0000)]
Update runtime shared memory computations.

23 years agoAdd SET SESSION AUTHORIZATION command.
Peter Eisentraut [Tue, 8 May 2001 21:06:43 +0000 (21:06 +0000)]
Add SET SESSION AUTHORIZATION command.

23 years agoMention SHELL usage in initdb.
Bruce Momjian [Tue, 8 May 2001 20:38:00 +0000 (20:38 +0000)]
Mention SHELL usage in initdb.

23 years agoAppend and SubqueryScan nodes were not passing changed-parameter signals down
Tom Lane [Tue, 8 May 2001 19:47:02 +0000 (19:47 +0000)]
Append and SubqueryScan nodes were not passing changed-parameter signals down
to their children, leading to misbehavior if they had any children that paid
attention to chgParam (most plan node types don't).  Append's bug has been
there a long time, but nobody had noticed because it used to be difficult
to create a query where an Append would be used below the top level of a
plan; so there were never any parameters getting passed down.  SubqueryScan
is new in 7.1 ... and I'd modeled its behavior on Append :-(

23 years agoNeed to factor out strdup.o for separate treatment since it's in a
Peter Eisentraut [Tue, 8 May 2001 19:38:57 +0000 (19:38 +0000)]
Need to factor out strdup.o for separate treatment since it's in a
different directory.  This makes dependency tracking work and copes with
compilers that don't suport -c and -o together.

23 years agoRemove pgadmin.sgml file.
Bruce Momjian [Tue, 8 May 2001 19:28:01 +0000 (19:28 +0000)]
Remove pgadmin.sgml file.

23 years agoAdd new emails to file.
Bruce Momjian [Tue, 8 May 2001 19:21:46 +0000 (19:21 +0000)]
Add new emails to file.

23 years agoRemove LISP manual reference. Added to web "interfaces" page.
Bruce Momjian [Tue, 8 May 2001 19:14:52 +0000 (19:14 +0000)]
Remove LISP manual reference.  Added to web "interfaces" page.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 8 May 2001 19:01:55 +0000 (19:01 +0000)]
Update TODO list.

23 years agoSmall cleanup.
Bruce Momjian [Tue, 8 May 2001 19:00:26 +0000 (19:00 +0000)]
Small cleanup.

23 years agodrop_function.sgml:
Bruce Momjian [Tue, 8 May 2001 17:51:30 +0000 (17:51 +0000)]
drop_function.sgml:
===================
In Notes:
 Refer to CREATE FUNCTION for information on creating aggregate functions.
                                                      ^^^^^^^^^^^^^^^^^^^
I assume it must read C function instead.

In Compatibility SQL/PSM:
 SQL/PSM is a proposed standard.

 We had that before: remove proposed.

drop_index.sgml:
================
<REFNAME>: Removes existing indexes from a database

as far as I can see index should be singular. The command description is
written as if only one index can be removed at a time. Interestingly
enough, in v7.0.2 it was in fact singular. Am I mistaken here?

drop_operator.sgml:
===================
In Outputs the arguments are referred to as type and type2, but the synopsis
and Inputs section these are left_type and right_type, respectively. Also,
oper is used in Outputs versus id in Inputs/Synopsis. In the translation I
follow the replaceables used in the Inputs/Synopsis part.

Frank Wegmann

23 years agodrop_function.sgml:
Bruce Momjian [Tue, 8 May 2001 17:35:57 +0000 (17:35 +0000)]
drop_function.sgml:
===================
In Notes:
 Refer to CREATE FUNCTION for information on creating aggregate functions.
                                                      ^^^^^^^^^^^^^^^^^^^
I assume it must read C function instead.

Frank Wegmann

23 years agoAdd newlines around debug output in optimizer showing total costs.
Bruce Momjian [Tue, 8 May 2001 17:25:28 +0000 (17:25 +0000)]
Add newlines around debug output in optimizer showing total costs.

23 years agoRun pgindent on ODBC code only, to reformat new comments.
Bruce Momjian [Tue, 8 May 2001 17:12:36 +0000 (17:12 +0000)]
Run pgindent on ODBC code only, to reformat new comments.

23 years agoODBC source code cleanup patch. Should match rest of PostgreSQL code better.
Bruce Momjian [Tue, 8 May 2001 17:06:18 +0000 (17:06 +0000)]
ODBC source code cleanup patch.  Should match rest of PostgreSQL code better.

23 years agoAllow SHELL in Makefile.* to control initdb.
Bruce Momjian [Tue, 8 May 2001 16:28:46 +0000 (16:28 +0000)]
Allow SHELL in Makefile.* to control initdb.

23 years agoUpdate readme about user registration.
Bruce Momjian [Tue, 8 May 2001 15:52:02 +0000 (15:52 +0000)]
Update readme about user registration.

23 years agoFix paren typo in java.
Bruce Momjian [Tue, 8 May 2001 15:29:14 +0000 (15:29 +0000)]
Fix paren typo in java.

23 years agoFurther cleanup of RAISE discussion; in particular, remove example
Tom Lane [Tue, 8 May 2001 02:53:24 +0000 (02:53 +0000)]
Further cleanup of RAISE discussion; in particular, remove example
that doesn't actually work.

23 years agoFix raise mention that %% output as %.
Bruce Momjian [Tue, 8 May 2001 02:26:09 +0000 (02:26 +0000)]
Fix raise mention that %% output as %.

23 years agoUn-break exec_move_row() for case that a NULL tuple and tupdesc are
Tom Lane [Tue, 8 May 2001 01:00:53 +0000 (01:00 +0000)]
Un-break exec_move_row() for case that a NULL tuple and tupdesc are
passed, which occurs when no rows are retrieved by a SELECT.
Mea maxima culpa ... I should have caught this.

23 years agoUpdate RAISE manual concerning % format symbols.
Bruce Momjian [Tue, 8 May 2001 00:30:41 +0000 (00:30 +0000)]
Update RAISE manual concerning % format symbols.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 8 May 2001 00:19:19 +0000 (00:19 +0000)]
Update TODO list.

23 years agoUpdate TODO list.
Bruce Momjian [Tue, 8 May 2001 00:14:15 +0000 (00:14 +0000)]
Update TODO list.

23 years agoUpdate RAISE mention in manual to take string/variable not format/identifier.
Bruce Momjian [Tue, 8 May 2001 00:09:22 +0000 (00:09 +0000)]
Update RAISE mention in manual to take string/variable not format/identifier.

23 years agoFix comment in file about 2*max_connections.
Bruce Momjian [Mon, 7 May 2001 23:32:55 +0000 (23:32 +0000)]
Fix comment in file about 2*max_connections.

23 years agoShared library and GCC support for SCO OpenServer.
Peter Eisentraut [Mon, 7 May 2001 20:43:28 +0000 (20:43 +0000)]
Shared library and GCC support for SCO OpenServer.

23 years agoAdd mention of functional indexes for case-insensitive comparisons.
Bruce Momjian [Mon, 7 May 2001 19:58:31 +0000 (19:58 +0000)]
Add mention of functional indexes for case-insensitive comparisons.

23 years agoFix remaining RI permission problems (cascaded update/delete, restrict,
Peter Eisentraut [Mon, 7 May 2001 19:57:24 +0000 (19:57 +0000)]
Fix remaining RI permission problems (cascaded update/delete, restrict,
set null/default).

23 years agoAdd \cd command to psql.
Peter Eisentraut [Mon, 7 May 2001 19:31:33 +0000 (19:31 +0000)]
Add \cd command to psql.

23 years agoUpdate TODO list.
Bruce Momjian [Mon, 7 May 2001 18:40:59 +0000 (18:40 +0000)]
Update TODO list.

23 years ago> Occasionally and without warning I get this from my daily vacuum
Bruce Momjian [Mon, 7 May 2001 17:20:19 +0000 (17:20 +0000)]
> Occasionally and without warning I get this from my daily vacuum
> cronjob:
> NOTICE:  RegisterSharedInvalid: SI buffer overflow
> NOTICE:  InvalidateSharedInvalid: cache state reset
> I don't understand what these mean. Should I be concerned about them
> and what do they signify?

No real need to worry.  Those should've been downgraded to DEBUG-level
messages a release or two back, but nobody bothered...

Tom Lane

23 years agoUpdate TODO list.
Bruce Momjian [Mon, 7 May 2001 17:05:17 +0000 (17:05 +0000)]
Update TODO list.

23 years agoUpdate FAQ to use new pgsql-cygwin@postgresql.org list.
Bruce Momjian [Mon, 7 May 2001 16:02:38 +0000 (16:02 +0000)]
Update FAQ to use new pgsql-cygwin@postgresql.org list.

23 years agoSolaris startup cleanup from Justin Clift
Bruce Momjian [Mon, 7 May 2001 15:55:27 +0000 (15:55 +0000)]
Solaris startup cleanup from Justin Clift

23 years agoRewrite of planner statistics-gathering code. ANALYZE is now available as
Tom Lane [Mon, 7 May 2001 00:43:27 +0000 (00:43 +0000)]
Rewrite of planner statistics-gathering code.  ANALYZE is now available as
a separate statement (though it can still be invoked as part of VACUUM, too).
pg_statistic redesigned to be more flexible about what statistics are
stored.  ANALYZE now collects a list of several of the most common values,
not just one, plus a histogram (not just the min and max values).  Random
sampling is used to make the process reasonably fast even on very large
tables.  The number of values and histogram bins collected is now
user-settable via an ALTER TABLE command.

There is more still to do; the new stats are not being used everywhere
they could be in the planner.  But the remaining changes for this project
should be localized, and the behavior is already better than before.

A not-very-related change is that sorting now makes use of btree comparison
routines if it can find one, rather than invoking '<' twice.

23 years agoFix collateral damage from previous (rev 1.49) patch.
Peter Eisentraut [Sun, 6 May 2001 21:15:51 +0000 (21:15 +0000)]
Fix collateral damage from previous (rev 1.49) patch.

23 years agoAllow 'psql -f -' to read from standard input.
Peter Eisentraut [Sun, 6 May 2001 17:38:32 +0000 (17:38 +0000)]
Allow 'psql -f -' to read from standard input.