]> granicus.if.org Git - postgresql/log
postgresql
21 years agoFix some problems with dropped columns in pltcl functions.
Tom Lane [Thu, 4 Sep 2003 15:10:10 +0000 (15:10 +0000)]
Fix some problems with dropped columns in pltcl functions.

21 years agoAllow non-super users to set log_duration to true, then false, when the
Bruce Momjian [Thu, 4 Sep 2003 05:11:20 +0000 (05:11 +0000)]
Allow non-super users to set log_duration to true, then false, when the
administrator has not turned it on, and fix other PGC_USERLIMIT
variables.

21 years agoRe-upper case CHECKPOINT_SEGMENTS.
Bruce Momjian [Thu, 4 Sep 2003 03:38:55 +0000 (03:38 +0000)]
Re-upper case CHECKPOINT_SEGMENTS.

21 years agoLowercase CHECKPOINT_SEGMENTS.
Bruce Momjian [Thu, 4 Sep 2003 00:36:46 +0000 (00:36 +0000)]
Lowercase CHECKPOINT_SEGMENTS.

21 years agoAdd:
Bruce Momjian [Wed, 3 Sep 2003 23:01:47 +0000 (23:01 +0000)]
Add:

> * Allow to_char to print localized month names (Karel)

21 years agoTighten casting of function call pointers for threads.
Bruce Momjian [Wed, 3 Sep 2003 22:34:08 +0000 (22:34 +0000)]
Tighten casting of function call pointers for threads.

21 years agoAdd note that a for loop where the lower bound is greater than the upper
Peter Eisentraut [Wed, 3 Sep 2003 22:17:07 +0000 (22:17 +0000)]
Add note that a for loop where the lower bound is greater than the upper
bound is valid but does nothing.

suggested by Richard Huxton <dev@archonet.com>

21 years agoPass session_authorization to the client and make psql update its prompt
Peter Eisentraut [Wed, 3 Sep 2003 22:05:09 +0000 (22:05 +0000)]
Pass session_authorization to the client and make psql update its prompt
accordingly.

21 years agoDone:
Bruce Momjian [Wed, 3 Sep 2003 21:27:21 +0000 (21:27 +0000)]
Done:

* -Have standalone backend read postgresql.conf (Tom)

21 years agoMark unixware as verified.
Bruce Momjian [Wed, 3 Sep 2003 20:54:21 +0000 (20:54 +0000)]
Mark unixware as verified.

21 years agoAdd Unixware as verified.
Bruce Momjian [Wed, 3 Sep 2003 20:51:40 +0000 (20:51 +0000)]
Add Unixware as verified.

21 years agoUpdate German FAQ.
Bruce Momjian [Wed, 3 Sep 2003 20:28:14 +0000 (20:28 +0000)]
Update German FAQ.

Ian Barwick

21 years agoMark BSD/OS as verified not needing reentrant function names.
Bruce Momjian [Wed, 3 Sep 2003 19:40:56 +0000 (19:40 +0000)]
Mark BSD/OS as verified not needing reentrant function names.

21 years agoAdd reporting of specificy threading failure functions.
Bruce Momjian [Wed, 3 Sep 2003 19:36:31 +0000 (19:36 +0000)]
Add reporting of specificy threading failure functions.

21 years agoAdd test for thread-safeness of libc functions.
Bruce Momjian [Wed, 3 Sep 2003 19:30:31 +0000 (19:30 +0000)]
Add test for thread-safeness of libc functions.

21 years agoMark Unixware as not requiring *_r functions.
Bruce Momjian [Wed, 3 Sep 2003 17:55:43 +0000 (17:55 +0000)]
Mark Unixware as not requiring *_r functions.

21 years agoWork around ENOTEMPTY being an alias for EEXIST on some platforms.
Tom Lane [Wed, 3 Sep 2003 15:49:42 +0000 (15:49 +0000)]
Work around ENOTEMPTY being an alias for EEXIST on some platforms.

21 years agoRepair problems with to_char() overrunning its input string.
Tom Lane [Wed, 3 Sep 2003 14:59:41 +0000 (14:59 +0000)]
Repair problems with to_char() overrunning its input string.
From Karel Zak.

21 years agoIn _bt_check_unique() loop, don't bother applying _bt_isequal() to
Tom Lane [Tue, 2 Sep 2003 22:10:16 +0000 (22:10 +0000)]
In _bt_check_unique() loop, don't bother applying _bt_isequal() to
killed items; just skip to the next item immediately.  Only check for
key equality when we reach a non-killed item or the end of the index
page.  This saves key comparisons when there are lots of killed items,
as for example in a heavily-updated table that's not been vacuumed lately.
Seems to be a win for pgbench anyway.

21 years agoRemove outdated CLI things.
Peter Eisentraut [Tue, 2 Sep 2003 21:45:37 +0000 (21:45 +0000)]
Remove outdated CLI things.

21 years agoCause standalone backend (including bootstrap case) to read the GUC
Tom Lane [Tue, 2 Sep 2003 19:04:12 +0000 (19:04 +0000)]
Cause standalone backend (including bootstrap case) to read the GUC
config file if it exists.  This was already discussed as being a good
idea, and now seems the cleanest way to deal with initdb-time failures
on machines with small SHMMAX.  (The submitted patches instead modified
initdb.sh to pass the correct sizing parameters, but that would still
leave standalone backends prone to failure later.  An admin who needs
to use a standalone backend has enough trouble already, he shouldn't
have to manually configure its shmem settings...)

21 years agoSeveral fixes for hash indexes that involve changing the on-disk index
Tom Lane [Tue, 2 Sep 2003 18:13:32 +0000 (18:13 +0000)]
Several fixes for hash indexes that involve changing the on-disk index
layout; therefore, this change forces REINDEX of hash indexes (though
not a full initdb).  Widen hashm_ntuples to double so that hash space
management doesn't get confused by more than 4G entries; enlarge the
allowed number of free-space-bitmap pages; replace the useless bshift
field with a useful bmshift field; eliminate 4 bytes of wasted space
in the per-page special area.

21 years agoFix a couple typos, add some more comments.
Tom Lane [Tue, 2 Sep 2003 03:29:01 +0000 (03:29 +0000)]
Fix a couple typos, add some more comments.

21 years agoRewrite hashbulkdelete() to make it amenable to new bucket locking
Tom Lane [Tue, 2 Sep 2003 02:18:38 +0000 (02:18 +0000)]
Rewrite hashbulkdelete() to make it amenable to new bucket locking
scheme.  A pleasant side effect is that it is *much* faster when deleting
a large fraction of the indexed tuples, because of elimination of
redundant hash_step activity induced by hash_adjscans.  Various other
continuing code cleanup.

21 years agoDo not pass server_encoding to the client.
Peter Eisentraut [Mon, 1 Sep 2003 23:04:49 +0000 (23:04 +0000)]
Do not pass server_encoding to the client.

libpq, talking to an old server, should assume SQL_ASCII as the default
client encoding, because that is what the server will actually use (not
the server encoding).

21 years agoClean up after pygresql removal: adjust/remove documentation and remove
Peter Eisentraut [Mon, 1 Sep 2003 23:01:49 +0000 (23:01 +0000)]
Clean up after pygresql removal: adjust/remove documentation and remove
unneeded configure work.

21 years agoPreliminary cleanup for hash index code (doesn't attack the locking problem
Tom Lane [Mon, 1 Sep 2003 20:26:34 +0000 (20:26 +0000)]
Preliminary cleanup for hash index code (doesn't attack the locking problem
yet).  Fix a couple of bugs that would only appear if multiple bitmap pages
are used, including a buffer reference leak and incorrect computation of bit
indexes.  Get rid of 'overflow address' concept, which accomplished nothing
except obfuscating the code and creating a risk of failure due to limited
range of offset field.  Rename some misleadingly-named fields and routines,
and improve documentation.

21 years agoAdd some internals documentation for hash indexes, including an
Tom Lane [Mon, 1 Sep 2003 20:24:49 +0000 (20:24 +0000)]
Add some internals documentation for hash indexes, including an
explanation of the remarkably confusing page addressing scheme.
The file also includes my planned-but-not-yet-implemented revision
of the hash index locking scheme.

21 years agoAdd mention of OID wraparound.
Bruce Momjian [Mon, 1 Sep 2003 16:47:58 +0000 (16:47 +0000)]
Add mention of OID wraparound.

21 years agoFixed two bugs in pgtypes library.
Michael Meskes [Mon, 1 Sep 2003 12:37:40 +0000 (12:37 +0000)]
Fixed two bugs in pgtypes library.

21 years agoRename DoIt to changeVar, for clarity.
Bruce Momjian [Mon, 1 Sep 2003 04:15:51 +0000 (04:15 +0000)]
Rename DoIt to changeVar, for clarity.

21 years agoTranslation update
Peter Eisentraut [Sun, 31 Aug 2003 17:38:06 +0000 (17:38 +0000)]
Translation update

21 years agoAdd/edit index entries.
Peter Eisentraut [Sun, 31 Aug 2003 17:32:24 +0000 (17:32 +0000)]
Add/edit index entries.

21 years agoSet LC_ALL=C when running collateindex.pl to get predictable sort order.
Peter Eisentraut [Sun, 31 Aug 2003 17:28:39 +0000 (17:28 +0000)]
Set LC_ALL=C when running collateindex.pl to get predictable sort order.

21 years agoGuard against pgindent changing =- to = -.
Bruce Momjian [Sat, 30 Aug 2003 14:59:34 +0000 (14:59 +0000)]
Guard against pgindent changing =- to = -.

21 years agoFix unixware template --- had spaces around equals.
Bruce Momjian [Sat, 30 Aug 2003 03:19:47 +0000 (03:19 +0000)]
Fix unixware template --- had spaces around equals.

21 years agoMake documentation of pg_restore's -N, -o, -r switches bear some
Tom Lane [Thu, 28 Aug 2003 20:44:02 +0000 (20:44 +0000)]
Make documentation of pg_restore's -N, -o, -r switches bear some
slight resemblance to their actual behavior.

21 years agopg_dump and pg_restore -r had managed to diverge on the ordering of
Tom Lane [Thu, 28 Aug 2003 20:21:34 +0000 (20:21 +0000)]
pg_dump and pg_restore -r had managed to diverge on the ordering of
different object types.  Fix, and centralize logic to try to prevent
the same mistake in future.

21 years agoEnsure that FK constraints are dumped after primary/unique constraints,
Tom Lane [Thu, 28 Aug 2003 18:59:06 +0000 (18:59 +0000)]
Ensure that FK constraints are dumped after primary/unique constraints,
rather than relying on OID order of pg_constraint entries.  Per gripe
from Greg Stark.

21 years agoFix stupid typo (mine I suppose) in CopyGetData. Per report from Dave Cramer.
Tom Lane [Thu, 28 Aug 2003 13:52:34 +0000 (13:52 +0000)]
Fix stupid typo (mine I suppose) in CopyGetData.  Per report from Dave Cramer.

21 years agoMore accuracy works with stopwords in queries
Teodor Sigaev [Thu, 28 Aug 2003 12:23:24 +0000 (12:23 +0000)]
More accuracy works with stopwords in queries

21 years agoAdjust date/time input parsing code to correctly distinguish the four
Tom Lane [Wed, 27 Aug 2003 23:29:29 +0000 (23:29 +0000)]
Adjust date/time input parsing code to correctly distinguish the four
SQLSTATE error codes required by SQL99 (invalid format, datetime field
overflow, interval field overflow, invalid time zone displacement value).
Also emit a HINT about DateStyle in cases where it seems appropriate.
Per recent gripes.

21 years agoAdd "override" for make variable JADEFLAGS, since upper directory passes
Peter Eisentraut [Wed, 27 Aug 2003 22:58:54 +0000 (22:58 +0000)]
Add "override" for make variable JADEFLAGS, since upper directory passes
this variable on the command line.

21 years agoMajor cleanup of SPI chapter
Peter Eisentraut [Wed, 27 Aug 2003 22:13:35 +0000 (22:13 +0000)]
Major cleanup of SPI chapter

21 years agoDepartment of second thoughts: can't reorder merge-join clauses.
Tom Lane [Wed, 27 Aug 2003 12:44:12 +0000 (12:44 +0000)]
Department of second thoughts: can't reorder merge-join clauses.

21 years agoUpdate URL for 7.3 changes:
Bruce Momjian [Wed, 27 Aug 2003 03:35:35 +0000 (03:35 +0000)]
Update URL for 7.3 changes:

http://developer.postgresql.org/~momjian/upgrade_tips_7.3

21 years agoupdate to beta2 REL7_4_BETA2
PostgreSQL Daemon [Wed, 27 Aug 2003 02:35:38 +0000 (02:35 +0000)]
update to beta2

21 years agoShare PG_DIAG_* macros between client and server and use them internally.
Peter Eisentraut [Wed, 27 Aug 2003 00:33:34 +0000 (00:33 +0000)]
Share PG_DIAG_* macros between client and server and use them internally.

21 years agoPush subplan clauses to the back in qual lists for join plans, not
Tom Lane [Tue, 26 Aug 2003 22:56:51 +0000 (22:56 +0000)]
Push subplan clauses to the back in qual lists for join plans, not
only scan plans.  Per observation from Rod Taylor.

21 years agoCall it Linux, not GNU/Linux.
Bruce Momjian [Tue, 26 Aug 2003 21:31:11 +0000 (21:31 +0000)]
Call it Linux, not GNU/Linux.

21 years agoUse separate SQLSTATE codes for file not found/file exists, rather than
Tom Lane [Tue, 26 Aug 2003 21:15:27 +0000 (21:15 +0000)]
Use separate SQLSTATE codes for file not found/file exists, rather than
lumping them into ERRCODE_UNDEFINED_OBJECT/ERRCODE_DUPLICATE_OBJECT.
This seems reasonable since 'object' was meant to refer to 'object in the
database' and a file is outside the database.  Per request from Dave
Cramer.

21 years agopsql's HISTSIZE from .psqlrc do not want to be set so I did it for psql
Bruce Momjian [Tue, 26 Aug 2003 18:35:31 +0000 (18:35 +0000)]
psql's HISTSIZE from .psqlrc do not want to be set so I did it for psql
that HISTSIZE might take effect from my .psqlrc

Andriy Tkachuk

21 years agoThis patch adds a new regression test for the UPDATE command. Right
Bruce Momjian [Tue, 26 Aug 2003 18:32:23 +0000 (18:32 +0000)]
This patch adds a new regression test for the UPDATE command. Right
now all that is tested is Rod Taylor's recent addition to allow
this syntax:

    UPDATE ... SET <col> = DEFAULT;

If anyone else would like to add more UPDATE tests, go ahead --
I just wanted to write a test for the above functionality, and
couldn't see an existing test that it would be appropriate
to add to.

Neil Conway

21 years agoFixed connect statement with username in variable.
Michael Meskes [Tue, 26 Aug 2003 16:09:02 +0000 (16:09 +0000)]
Fixed connect statement with username in variable.

21 years agoAttached is a (very small) patch to make analyze display some
Bruce Momjian [Tue, 26 Aug 2003 15:38:42 +0000 (15:38 +0000)]
Attached is a (very small) patch to make analyze display some
interesting info in verbose mode about the analyzed relation (pages,
rows per block and rows).

Mark Kirkwood

21 years agoMop-up for previous change to determine default shared_buffers and
Tom Lane [Tue, 26 Aug 2003 15:38:25 +0000 (15:38 +0000)]
Mop-up for previous change to determine default shared_buffers and
max_connections at initdb time.  Get rid of DEF_NBUFFERS and DEF_MAXBACKENDS
macros, which aren't doing anything useful anymore, and put more likely
defaults into postgresql.conf.sample.

21 years agofix module name in MODULE_PATHNAME substitution
Teodor Sigaev [Tue, 26 Aug 2003 10:01:02 +0000 (10:01 +0000)]
fix module name in MODULE_PATHNAME substitution

21 years agoAttempt to fix setMaxFieldSize() logic that was checked in yesterday.
Barry Lind [Tue, 26 Aug 2003 06:50:39 +0000 (06:50 +0000)]
Attempt to fix setMaxFieldSize() logic that was checked in yesterday.
I think this should fix the problem, but since I don't have a reproducable test
case, I can't be sure.  This problem is reported by Kim Ho of redhat, who will
test this fix.  This also includes a test case for the original functionality.

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java
  jdbc/org/postgresql/test/jdbc2/ResultSetTest.java

21 years agoAdd the Brazilian time zone abbreviations BRT, BRST, FNT, FNST.
Tom Lane [Mon, 25 Aug 2003 23:30:27 +0000 (23:30 +0000)]
Add the Brazilian time zone abbreviations BRT, BRST, FNT, FNST.
ACT and ACST were already present.  AMT and AMST conflict with the
existing entries for Armenia; no change there for the moment.

21 years agoAllow parsing of time and timetz inputs to accept the documented input
Tom Lane [Mon, 25 Aug 2003 22:47:34 +0000 (22:47 +0000)]
Allow parsing of time and timetz inputs to accept the documented input
syntax '040506' for '04:05:06', as well as '0405' for '04:05:00'.  This
has been broken since 7.2 but was only recently complained of.

21 years agoTranslation updates
Peter Eisentraut [Mon, 25 Aug 2003 21:29:53 +0000 (21:29 +0000)]
Translation updates

21 years agoRefactor code so that to_date() does not call to_timestamp() and then
Tom Lane [Mon, 25 Aug 2003 16:13:27 +0000 (16:13 +0000)]
Refactor code so that to_date() does not call to_timestamp() and then
perform a timestamp-to-date coercion.  Instead both routines share a
subroutine that delivers the parsing result as a struct tm.  This avoids
problems with timezone dependency of to_date's result, and should be
at least marginally faster too.

21 years agoSynced parser and fixed a bug in error output to log file.
Michael Meskes [Mon, 25 Aug 2003 13:44:00 +0000 (13:44 +0000)]
Synced parser and fixed a bug in error output to log file.

21 years agoAdd:
Bruce Momjian [Mon, 25 Aug 2003 04:10:59 +0000 (04:10 +0000)]
Add:

> * Add a libpq function to support Parse/DescribeStatement capability

21 years agoFix unixware thread test.
Bruce Momjian [Mon, 25 Aug 2003 02:48:25 +0000 (02:48 +0000)]
Fix unixware thread test.

21 years agoFix GB18030 to UTF-8 mapping table
Tatsuo Ishii [Mon, 25 Aug 2003 01:46:16 +0000 (01:46 +0000)]
Fix GB18030 to UTF-8 mapping table

21 years agoAdd quotes to += of unixware threads define.
Bruce Momjian [Sun, 24 Aug 2003 23:21:05 +0000 (23:21 +0000)]
Add quotes to += of unixware threads define.

21 years agoAdd:
Bruce Momjian [Sun, 24 Aug 2003 22:57:21 +0000 (22:57 +0000)]
Add:

* Allow TRUNCATE ... CASCADE/RESTRICT

21 years agoApplied patches from Oliver Jowett to fix the following bugs:
Barry Lind [Sun, 24 Aug 2003 22:10:09 +0000 (22:10 +0000)]
Applied patches from Oliver Jowett to fix the following bugs:
 - adds a finalizer method to AbstractJdbc1Statement to clean up in the case of
   poor user code which fails to close the statement object
 - fix ant build file to correctly detect dependencies across jdbc1/jdbc2/jdbc3
 - fix a coupld of server prepared statement bugs and added regression test for
   them
Applied patch from Kim Ho:
 - adds support for get/setMaxFieldSize().
Also fixed build.xml to provide a better error message in the event that an
older version of the driver exists in the classpath when trying to build.

21 years agoFix translation mistake.
Peter Eisentraut [Sun, 24 Aug 2003 21:26:34 +0000 (21:26 +0000)]
Fix translation mistake.

21 years agoTranslation updates
Peter Eisentraut [Sun, 24 Aug 2003 21:18:53 +0000 (21:18 +0000)]
Translation updates

21 years agoChange warnings for non-existing or pre-existing cursors to errors.
Peter Eisentraut [Sun, 24 Aug 2003 21:02:43 +0000 (21:02 +0000)]
Change warnings for non-existing or pre-existing cursors to errors.

21 years agoAdd macros for error result fields to libpq.
Peter Eisentraut [Sun, 24 Aug 2003 18:36:38 +0000 (18:36 +0000)]
Add macros for error result fields to libpq.

21 years agoFix GB18030 to UTF-8 mapping table
Tatsuo Ishii [Sun, 24 Aug 2003 05:18:04 +0000 (05:18 +0000)]
Fix GB18030 to UTF-8 mapping table

21 years agoBring the makefiles up to our conventions.
Peter Eisentraut [Sat, 23 Aug 2003 04:25:29 +0000 (04:25 +0000)]
Bring the makefiles up to our conventions.

21 years agoEnable uninstalling data files that were installed from a subdirectory of
Peter Eisentraut [Sat, 23 Aug 2003 04:23:38 +0000 (04:23 +0000)]
Enable uninstalling data files that were installed from a subdirectory of
the source tree.

21 years agoFix uninstall target.
Peter Eisentraut [Sat, 23 Aug 2003 04:22:34 +0000 (04:22 +0000)]
Fix uninstall target.

21 years agoCorrect uninstall target.
Peter Eisentraut [Sat, 23 Aug 2003 04:21:59 +0000 (04:21 +0000)]
Correct uninstall target.

21 years agoTweak grammar to use FastAppend rather than lappend when constructing
Tom Lane [Fri, 22 Aug 2003 20:34:33 +0000 (20:34 +0000)]
Tweak grammar to use FastAppend rather than lappend when constructing
expr_lists.  This appears to be the only remaining O(N^2) bottleneck
in processing many-way 'x IN (a,b,c,...)' conditions.

21 years agoTweak processing of multiple-index-scan plans to reduce overhead when
Tom Lane [Fri, 22 Aug 2003 20:26:43 +0000 (20:26 +0000)]
Tweak processing of multiple-index-scan plans to reduce overhead when
handling many-way scans: instead of re-evaluating all prior indexscan
quals to see if a tuple has been fetched more than once, use a hash table
indexed by tuple CTID.  But fall back to the old way if the hash table
grows to exceed SortMem.

21 years agoISpell info updated
Teodor Sigaev [Thu, 21 Aug 2003 15:53:21 +0000 (15:53 +0000)]
ISpell info updated

21 years agoTranslation updates
Peter Eisentraut [Wed, 20 Aug 2003 21:10:01 +0000 (21:10 +0000)]
Translation updates

21 years agoUpdates for array documentation, from Joe Conway.
Tom Lane [Tue, 19 Aug 2003 06:06:48 +0000 (06:06 +0000)]
Updates for array documentation, from Joe Conway.

21 years agoImprove dynahash.c's API so that caller can specify the comparison function
Tom Lane [Tue, 19 Aug 2003 01:13:41 +0000 (01:13 +0000)]
Improve dynahash.c's API so that caller can specify the comparison function
as well as the hash function (formerly the comparison function was hardwired
as memcmp()).  This makes it possible to eliminate the special-purpose
hashtable management code in execGrouping.c in favor of using dynahash to
manage tuple hashtables; which is a win because dynahash knows how to expand
a hashtable when the original size estimate was too small, whereas the
special-purpose code was too stupid to do that.  (See recent gripe from
Stephan Szabo about poor performance when hash table size estimate is way
off.)  Free side benefit: when using string_hash, the default comparison
function is now strncmp() instead of memcmp().  This should eliminate some
part of the overhead associated with larger NAMEDATALEN values.

21 years agoWhen compiling a plpgsql trigger function, include the OID of the table
Tom Lane [Mon, 18 Aug 2003 19:16:02 +0000 (19:16 +0000)]
When compiling a plpgsql trigger function, include the OID of the table
the trigger is attached to in the hashkey.  This ensures that we will
create separate compiled trees for each table the trigger is used with,
avoiding possible datatype-mismatch problems if the tables have different
rowtypes.  This is essentially the same bug recently identified in plpython
--- though plpgsql doesn't seem as prone to crash when the rowtype changes
underneath it.  But failing robustly is no substitute for just working.

21 years agoDone:
Bruce Momjian [Mon, 18 Aug 2003 02:52:45 +0000 (02:52 +0000)]
Done:

> * -Add btree index support for reltime, tinterval, regproc (Tom)

21 years agoFix ARRAY[] construct so that in multidimensional case, elements can
Tom Lane [Sun, 17 Aug 2003 23:43:27 +0000 (23:43 +0000)]
Fix ARRAY[] construct so that in multidimensional case, elements can
be anything yielding an array of the proper kind, not only sub-ARRAY[]
constructs; do subscript checking at runtime not parse time.  Also,
adjust array_cat to make array || array comply with the SQL99 spec.

Joe Conway

21 years agoClean up locktable init code per recent gripe from Kurt Roeckx.
Tom Lane [Sun, 17 Aug 2003 22:41:12 +0000 (22:41 +0000)]
Clean up locktable init code per recent gripe from Kurt Roeckx.
No change in behavior, but old code would have failed to detect
overrun of MAX_LOCKMODES.

21 years agoSomebody forgot to include any actual documentation for ADD_MISSING_FROM.
Tom Lane [Sun, 17 Aug 2003 22:19:15 +0000 (22:19 +0000)]
Somebody forgot to include any actual documentation for ADD_MISSING_FROM.

21 years agoUpdate documentation to reflect the fact that ORDER BY, GROUP BY, etc
Tom Lane [Sun, 17 Aug 2003 22:09:00 +0000 (22:09 +0000)]
Update documentation to reflect the fact that ORDER BY, GROUP BY, etc
are now driven by the default btree opclass, rather than assuming that
particular operator names have the needed semantics.

21 years agoFix broken markup.
Tom Lane [Sun, 17 Aug 2003 22:05:13 +0000 (22:05 +0000)]
Fix broken markup.

21 years agoCreate a 'type cache' that keeps track of the data needed for any particular
Tom Lane [Sun, 17 Aug 2003 19:58:06 +0000 (19:58 +0000)]
Create a 'type cache'  that keeps track of the data needed for any particular
datatype by array_eq and array_cmp; use this to solve problems with memory
leaks in array indexing support.  The parser's equality_oper and ordering_oper
routines also use the cache.  Change the operator search algorithms to look
for appropriate btree or hash index opclasses, instead of assuming operators
named '<' or '=' have the right semantics.  (ORDER BY ASC/DESC now also look
at opclasses, instead of assuming '<' and '>' are the right things.)  Add
several more index opclasses so that there is no regression in functionality
for base datatypes.  initdb forced due to catalog additions.

21 years agoAdd to release notes.
Bruce Momjian [Sun, 17 Aug 2003 05:24:52 +0000 (05:24 +0000)]
Add to release notes.

21 years agoAdd:
Bruce Momjian [Sun, 17 Aug 2003 05:23:13 +0000 (05:23 +0000)]
Add:

> * Allow PREPARE of cursors

21 years agoI almost forgot mark in docs "to_char(interval)" as deprecated function.
Bruce Momjian [Sun, 17 Aug 2003 04:52:41 +0000 (04:52 +0000)]
I almost forgot mark in docs "to_char(interval)" as deprecated function.
This useless routine will removed in 7.5. It's already discussed (see
hackers list archive).

Karel Zak

21 years agoAdd:
Bruce Momjian [Sun, 17 Aug 2003 04:50:57 +0000 (04:50 +0000)]
Add:

> * Fix upper()/lower() to work for multibyte encodings
>

21 years agoThis patch makes two minor fixes to the docs: (1) fixes a
Bruce Momjian [Sun, 17 Aug 2003 04:46:59 +0000 (04:46 +0000)]
This patch makes two minor fixes to the docs: (1) fixes a
spelling mistake in the PREPARE ref page (2) Makes some
English more consistent, in the ref pages for some of the
client apps (3) Adds a link to the libpq docs in the
vacuumdb ref page.

Neil Conway

21 years agoThere is a misstatement in the CLOSE reference page, now that we
Bruce Momjian [Sun, 17 Aug 2003 04:46:00 +0000 (04:46 +0000)]
There is a misstatement in the CLOSE reference page, now that we
have cursors that might outlive their creating transactions. A
patch is attached that fixes this (suggestions on better wording
are welcome).

Neil Conway

21 years agoAdd:
Bruce Momjian [Sun, 17 Aug 2003 04:43:48 +0000 (04:43 +0000)]
Add:

> * Allow pooled connections to query prepared queries
> * Allow pooled connections to close all open WITH HOLD cursors

21 years agoI just noticed that the sample pg_hba.conf in the docs doesn't contain
Bruce Momjian [Sun, 17 Aug 2003 04:39:11 +0000 (04:39 +0000)]
I just noticed that the sample pg_hba.conf in the docs doesn't contain
any use of CIDR masks - here's a patch that adds  a couple of  sample
lines and associated comments.

Andrew Dunstan