]> granicus.if.org Git - postgresql/log
postgresql
17 years agoFix a typo.
Neil Conway [Sat, 7 Oct 2006 03:17:13 +0000 (03:17 +0000)]
Fix a typo.

17 years agoAdd main documentation link tags to the release notes.
Bruce Momjian [Sat, 7 Oct 2006 02:18:34 +0000 (02:18 +0000)]
Add main documentation link tags to the release notes.

17 years agoFix string_to_array() to correctly handle the case where there are
Tom Lane [Sat, 7 Oct 2006 00:11:53 +0000 (00:11 +0000)]
Fix string_to_array() to correctly handle the case where there are
overlapping possible matches for the separator string, such as
string_to_array('123xx456xxx789', 'xx').
Also, revise the logic of replace(), split_part(), and string_to_array()
to avoid O(N^2) work from redundant searches and conversions to pg_wchar
format when there are N matches to the separator string.
Backpatched the full patch as far as 8.0.  7.4 also has the bug, but the
code has diverged a lot, so I just went for a quick-and-dirty fix of the
bug itself in that branch.

17 years agoFix bcc32.mak to create and remove pg_config_os.h properly. Apparently
Tom Lane [Fri, 6 Oct 2006 18:53:53 +0000 (18:53 +0000)]
Fix bcc32.mak to create and remove pg_config_os.h properly.  Apparently
win32.mak got patched for this, but not bcc32.mak.

17 years agoFix SysCacheGetAttr() to handle the case where the specified syscache has not
Tom Lane [Fri, 6 Oct 2006 18:23:35 +0000 (18:23 +0000)]
Fix SysCacheGetAttr() to handle the case where the specified syscache has not
been initialized yet.  This can happen because there are code paths that call
SysCacheGetAttr() on a tuple originally fetched from a different syscache
(hopefully on the same catalog) than the one specified in the call.  It
doesn't seem useful or robust to try to prevent that from happening, so just
improve the function to cope instead.  Per bug#2678 from Jeff Trout.  The
specific example shown by Jeff is new in 8.1, but to be on the safe side
I'm backpatching 8.0 as well.  We could patch 7.x similarly but I think
that's probably overkill, given the lack of evidence of old bugs of this ilk.

17 years agoMessage style improvements
Peter Eisentraut [Fri, 6 Oct 2006 17:14:01 +0000 (17:14 +0000)]
Message style improvements

17 years agoCleanup for pglz_compress code: remove dead code, const-ify API of
Tom Lane [Thu, 5 Oct 2006 23:33:33 +0000 (23:33 +0000)]
Cleanup for pglz_compress code: remove dead code, const-ify API of
remaining functions, simplify pglz_compress's API to not require a useless
data copy when compression fails.  Also add a check in pglz_decompress that
the expected amount of data was decompressed.

17 years agoMake use of qsort_arg in several places that were formerly using klugy
Tom Lane [Thu, 5 Oct 2006 17:57:40 +0000 (17:57 +0000)]
Make use of qsort_arg in several places that were formerly using klugy
static variables.  This avoids any risk of potential non-reentrancy,
and in particular offers a much cleaner workaround for the Intel compiler
bug that was affecting ginutil.c.

17 years agoAdd indenting to Solaris FAQ, for clarity.
Bruce Momjian [Thu, 5 Oct 2006 03:13:15 +0000 (03:13 +0000)]
Add indenting to Solaris FAQ, for clarity.

17 years agoUpdate DB2X texinfo example code.
Bruce Momjian [Thu, 5 Oct 2006 02:04:32 +0000 (02:04 +0000)]
Update DB2X texinfo example code.

Andreas Seltenreich

17 years agoWork around buggy strtod on (some versions of?) IRIX. Combination of
Tom Lane [Thu, 5 Oct 2006 01:40:45 +0000 (01:40 +0000)]
Work around buggy strtod on (some versions of?) IRIX.  Combination of
proposed patches from John Jorgensen and Steve Singer.

17 years agoSecond try at fixing libedit configuration for Bob Friesenhahn.
Tom Lane [Thu, 5 Oct 2006 00:07:45 +0000 (00:07 +0000)]
Second try at fixing libedit configuration for Bob Friesenhahn.
Buildfarm results from 'gazelle' show that there are indeed libedit
versions for which history.h is a needed header, even though it's
apparently been dropped entirely in other versions.  Grumble.

17 years agoUpdate Solaris strtod() bug comment.
Bruce Momjian [Wed, 4 Oct 2006 22:49:44 +0000 (22:49 +0000)]
Update Solaris strtod() bug comment.

17 years agoUpdate Solaris strtol() bug comment.
Bruce Momjian [Wed, 4 Oct 2006 22:48:41 +0000 (22:48 +0000)]
Update Solaris strtol() bug comment.

17 years agoAdjust configure's handling of libedit to not look for a history.h file.
Tom Lane [Wed, 4 Oct 2006 22:31:14 +0000 (22:31 +0000)]
Adjust configure's handling of libedit to not look for a history.h file.
Per Bob Friesenhahn's report, this file is not supplied by some versions
of libedit, and even when it is supplied it seems to be just a link to
readline.h, so we don't need to include it anyway.
Also, ensure that we won't try to use a too-old version of Bison.
The previous coding would bleat but then use it anyway; better to invoke
the 'missing' script if any grammar files need to be rebuilt.

17 years agoUpdate Solaris FAQ.
Bruce Momjian [Wed, 4 Oct 2006 22:03:22 +0000 (22:03 +0000)]
Update Solaris FAQ.

Zdenek Kotala

17 years agoExclude pgindent from affecting the ecpg regression directory.
Bruce Momjian [Wed, 4 Oct 2006 20:42:19 +0000 (20:42 +0000)]
Exclude pgindent from affecting the ecpg regression directory.

17 years agoFix initdb's logic for picking shared_buffers/max_fsm_pages, which had
Tom Lane [Wed, 4 Oct 2006 18:58:08 +0000 (18:58 +0000)]
Fix initdb's logic for picking shared_buffers/max_fsm_pages, which had
gotten rather thoroughly whacked out by careless recent changes: the
intended ratio between the two was off by a lot, and the minimum number
of shared buffers tried had increased by a lot.  Problem exposed by
failures on buildfarm members with smaller SHMMAX values.

17 years agoFix comment to agree with presentation.
Tom Lane [Wed, 4 Oct 2006 18:45:52 +0000 (18:45 +0000)]
Fix comment to agree with presentation.

17 years agoImprove error messages from to_tsquery per yesterday's discussion:
Tom Lane [Wed, 4 Oct 2006 17:52:52 +0000 (17:52 +0000)]
Improve error messages from to_tsquery per yesterday's discussion:
provide the bad input, and be sure to mention that we are talking about
a tsearch query.

17 years agoAdd qsort and qsort_arg to MSVC list of src/port files.
Tom Lane [Wed, 4 Oct 2006 16:51:22 +0000 (16:51 +0000)]
Add qsort and qsort_arg to MSVC list of src/port files.

17 years agoReverted changes made by pgindent
Michael Meskes [Wed, 4 Oct 2006 10:09:03 +0000 (10:09 +0000)]
Reverted changes made by pgindent
Made show.pgc set all variables before displaying them.

17 years agopgindent run for 8.2.
Bruce Momjian [Wed, 4 Oct 2006 00:30:14 +0000 (00:30 +0000)]
pgindent run for 8.2.

17 years agoUdpate typedefs for pgindent.
Bruce Momjian [Wed, 4 Oct 2006 00:02:10 +0000 (00:02 +0000)]
Udpate typedefs for pgindent.

17 years agoSwitch over to using our own qsort() all the time, as has been proposed
Tom Lane [Tue, 3 Oct 2006 22:18:23 +0000 (22:18 +0000)]
Switch over to using our own qsort() all the time, as has been proposed
repeatedly.  Now that we don't have to worry about memory leaks from
glibc's qsort, we can safely put CHECK_FOR_INTERRUPTS into the tuplesort
comparators, as was requested a couple months ago.  Also, get rid of
non-reentrancy and an extra level of function call in tuplesort.c by
providing a variant qsort_arg() API that passes an extra void * argument
through to the comparison routine.  (We might want to use that in other
places too, I didn't look yet.)

17 years agopgevent fixes:
Bruce Momjian [Tue, 3 Oct 2006 22:12:14 +0000 (22:12 +0000)]
pgevent fixes:

1) Make vcbuild actually build the pgevent dll.
2) Change the pgevent DLL file so it doens't specify ordinal for the
functions. You're not supposed to do that. You're actually supposed to
declare them as PRIVATE as well, but mingw doesn't support that. VC++
will throw a warning and not an error though, so we can live with it.

Magnus Hagander

17 years agoUpdate typedef list for 8.2 pgindent run.
Bruce Momjian [Tue, 3 Oct 2006 22:09:42 +0000 (22:09 +0000)]
Update typedef list for 8.2 pgindent run.

17 years agoFix yesno_prompt() memory leak in /script tools, reported by Converity.
Bruce Momjian [Tue, 3 Oct 2006 21:45:20 +0000 (21:45 +0000)]
Fix yesno_prompt() memory leak in /script tools, reported by Converity.

17 years agoImprove numeric overflow error message.
Bruce Momjian [Tue, 3 Oct 2006 21:25:56 +0000 (21:25 +0000)]
Improve numeric overflow error message.

David Fetter

17 years agoMake some sentences consistent with similar ones.
Bruce Momjian [Tue, 3 Oct 2006 21:21:36 +0000 (21:21 +0000)]
Make some sentences consistent with similar ones.

Euler Taveira de Oliveira

17 years agoImprove readline psql prompt suggestion.
Bruce Momjian [Tue, 3 Oct 2006 21:14:46 +0000 (21:14 +0000)]
Improve readline psql prompt suggestion.

Andreas Seltenreich

17 years agoAdd experimental support for db2x texinfo
Bruce Momjian [Tue, 3 Oct 2006 21:13:17 +0000 (21:13 +0000)]
Add experimental support for db2x texinfo

Andreas Seltenreich

17 years agoThe attached patch changes units of the some default values in
Bruce Momjian [Tue, 3 Oct 2006 21:11:55 +0000 (21:11 +0000)]
The attached patch changes units of the some default values in
postgresql.conf.

- shared_buffers = 32000kB => 32MB
- temp_buffers = 8000kB => 8MB
- wal_buffers = 8 => 64kB

The code of initdb was a bit modified to write MB-unit values.
Values greater than 8000kB are rounded out to MB.

GUC_UNIT_XBLOCKS is added for wal_buffers. It is like GUC_UNIT_BLOCKS,
but uses XLOG_BLCKSZ instead of BLCKSZ.

Also, I cleaned up the test of GUC_UNIT_* flags in preparation to
add more unit flags in less bits.

ITAGAKI Takahiro

17 years agoCleanup pgwin32_open() 'if' test, and avoid possible error.
Bruce Momjian [Tue, 3 Oct 2006 20:44:18 +0000 (20:44 +0000)]
Cleanup pgwin32_open() 'if' test, and avoid possible error.

17 years agoRearrange MSVC errcode hack, fix incorrect _MSC_VER test. Magnus
Tom Lane [Tue, 3 Oct 2006 20:33:20 +0000 (20:33 +0000)]
Rearrange MSVC errcode hack, fix incorrect _MSC_VER test.  Magnus

17 years agoMove WIN32_ONLY_COMPILER define from c.h to win32.h because it was being
Bruce Momjian [Tue, 3 Oct 2006 03:59:22 +0000 (03:59 +0000)]
Move WIN32_ONLY_COMPILER define from c.h to win32.h because it was being
defined too late.

17 years agoThis patch fixes a small typo in information_schema.sgml.
Bruce Momjian [Tue, 3 Oct 2006 01:03:53 +0000 (01:03 +0000)]
This patch fixes a small typo in information_schema.sgml.

Guillaume Lelarge

17 years agoMention in postgresql.conf that autovacuum also needs
Bruce Momjian [Tue, 3 Oct 2006 00:39:26 +0000 (00:39 +0000)]
Mention in postgresql.conf that autovacuum also needs
stats_start_collector and stats_row_level to also be on

David Wheeler

17 years agoAdd URL for strlcpy():
Bruce Momjian [Tue, 3 Oct 2006 00:03:54 +0000 (00:03 +0000)]
Add URL for strlcpy():

>
>   http://archives.postgresql.org/pgsql-hackers/2006-09/msg02108.php

17 years agoAdd:
Bruce Momjian [Tue, 3 Oct 2006 00:00:18 +0000 (00:00 +0000)]
Add:

>  * Fix SSL retry to avoid useless repeated connection attempts and
>    ensuing misleading error messages

17 years agoAdd URL about strlcpy() creation history.
Bruce Momjian [Mon, 2 Oct 2006 23:58:59 +0000 (23:58 +0000)]
Add URL about strlcpy() creation history.

17 years agoUpdate wording:
Bruce Momjian [Mon, 2 Oct 2006 23:26:02 +0000 (23:26 +0000)]
Update wording:

< * Use strlcpy() rather than StrNCpy() macro
> * Use strlcpy() rather than our StrNCpy() macro

17 years agoAdd:
Bruce Momjian [Mon, 2 Oct 2006 23:23:32 +0000 (23:23 +0000)]
Add:

>
> * Use strlcpy() rather than StrNCpy() macro

17 years agoRemove mention of the -fast option from the Solaris FAQ.
Bruce Momjian [Mon, 2 Oct 2006 23:01:17 +0000 (23:01 +0000)]
Remove mention of the -fast option from the Solaris FAQ.

Kenneth Marshall

17 years agoUpdate recovery documentation.
Bruce Momjian [Mon, 2 Oct 2006 22:33:02 +0000 (22:33 +0000)]
Update recovery documentation.

Simon Riggs

17 years agoUpdate tsearch2 README.
Bruce Momjian [Mon, 2 Oct 2006 22:32:10 +0000 (22:32 +0000)]
Update tsearch2 README.

Robert Treat

17 years agoAdd URL for cached copy of paper as SGML comment.
Bruce Momjian [Mon, 2 Oct 2006 22:28:48 +0000 (22:28 +0000)]
Add URL for cached copy of paper as SGML comment.

17 years agoAdd release note item:
Bruce Momjian [Mon, 2 Oct 2006 15:32:03 +0000 (15:32 +0000)]
Add release note item:

        Drop privileges on startup so servers can be started from
        an administrative account (Magnus)

17 years agoMove "Exotic Features" to the bottom of TODO.
Bruce Momjian [Mon, 2 Oct 2006 14:00:04 +0000 (14:00 +0000)]
Move "Exotic Features" to the bottom of TODO.

17 years agoProperly handle the case where strlcpy() exists in libc but isn't
Tom Lane [Mon, 2 Oct 2006 00:06:18 +0000 (00:06 +0000)]
Properly handle the case where strlcpy() exists in libc but isn't
declared in the system headers.  Per report from Bruce than some BSDen
are like this.

17 years agoImprove documentation of configure's readline/libedit switches.
Tom Lane [Sun, 1 Oct 2006 23:47:16 +0000 (23:47 +0000)]
Improve documentation of configure's readline/libedit switches.

17 years agoMake some marginal performance improvements in reportErrorPosition(),
Tom Lane [Sun, 1 Oct 2006 22:25:48 +0000 (22:25 +0000)]
Make some marginal performance improvements in reportErrorPosition(),
which turns out to be a dominant part of the runtime in scenarios
involving lots of parse-time warnings (such as Stephen Frost's example
of an INSERT with a lot of backslash-containing strings).  There's not
a whole lot we can do about the character-at-a-time scanning, but we
can at least avoid traversing the query twice.

17 years agoDo a CHECK_FOR_INTERRUPTS after emitting a message of less than ERROR
Tom Lane [Sun, 1 Oct 2006 22:08:18 +0000 (22:08 +0000)]
Do a CHECK_FOR_INTERRUPTS after emitting a message of less than ERROR
severity.  This is to ensure the user can cancel a query that's spitting
out lots of notice/warning messages, even if they're coming from a loop
that doesn't otherwise contain a CHECK_FOR_INTERRUPTS.  Per gripe from
Stephen Frost.

17 years agodate_trunc also accepts 'quarter'. Noted by Yoshihisa Nakano.
Tom Lane [Sun, 1 Oct 2006 18:54:31 +0000 (18:54 +0000)]
date_trunc also accepts 'quarter'.  Noted by Yoshihisa Nakano.

17 years agoRemove accented characters in comments, to avoid failures when this
Tom Lane [Sun, 1 Oct 2006 18:37:53 +0000 (18:37 +0000)]
Remove accented characters in comments, to avoid failures when this
file is read with an incompatible client_encoding setting.  Per report
from Tim N. van der Leeuw.

17 years agoFix overly enthusiastic Assert introduced in 8.1: it's expecting a
Tom Lane [Sun, 1 Oct 2006 17:23:38 +0000 (17:23 +0000)]
Fix overly enthusiastic Assert introduced in 8.1: it's expecting a
CaseTestExpr, but forgot that the optimizer is sometimes able to replace
CaseTestExpr by Const.

17 years agoSuppress compiler warnings.
Tom Lane [Sat, 30 Sep 2006 18:44:37 +0000 (18:44 +0000)]
Suppress compiler warnings.

17 years agouninstall script for sslinfo
Tom Lane [Sat, 30 Sep 2006 18:24:19 +0000 (18:24 +0000)]
uninstall script for sslinfo

17 years agouninstall script for pg_freespacemap
Tom Lane [Sat, 30 Sep 2006 18:15:48 +0000 (18:15 +0000)]
uninstall script for pg_freespacemap

17 years agouninstall script for pgrowlocks - Josh Drake
Tom Lane [Sat, 30 Sep 2006 18:08:32 +0000 (18:08 +0000)]
uninstall script for pgrowlocks - Josh Drake

17 years agoAdd uninstall script for adminpack - Josh Drake
Tom Lane [Sat, 30 Sep 2006 18:03:06 +0000 (18:03 +0000)]
Add uninstall script for adminpack - Josh Drake

17 years agoBSD/OS doesn't have a prototype for strlcpy() but has the function, so
Bruce Momjian [Sat, 30 Sep 2006 12:06:42 +0000 (12:06 +0000)]
BSD/OS doesn't have a prototype for strlcpy() but has the function, so
work around that with defines.

17 years agoAdd:
Bruce Momjian [Sat, 30 Sep 2006 03:55:32 +0000 (03:55 +0000)]
Add:

> * Allow more complex user/database default GUC settings
>   Currently, ALTER USER and ALTER DATABASE support per-user and
>   per-database defaults.  Consider adding per-user-and-database
>   defaults so things like search_path can be defaulted for a
>   specific user connecting to a specific database.
>
>

17 years agoAdd MONEY URL:
Bruce Momjian [Sat, 30 Sep 2006 02:59:06 +0000 (02:59 +0000)]
Add MONEY URL:

>   http://archives.postgresql.org/pgsql-hackers/2006-09/msg01107.php

17 years agoAllow assignment to array elements not contiguous with those already
Tom Lane [Fri, 29 Sep 2006 21:22:21 +0000 (21:22 +0000)]
Allow assignment to array elements not contiguous with those already
present; intervening positions are filled with nulls.  This behavior
is required by SQL99 but was not implementable before 8.2 due to lack
of support for nulls in arrays.  I have only made it work for the
one-dimensional case, which is all that SQL99 requires.  It seems quite
complex to get it right in higher dimensions, and since we never allowed
extension at all in higher dimensions, I think that must count as a
future feature addition not a bug fix.

17 years agoRemove duplicate (and now incorrect) documentation for row-wise IS [NOT]
Tom Lane [Fri, 29 Sep 2006 18:36:45 +0000 (18:36 +0000)]
Remove duplicate (and now incorrect) documentation for row-wise IS [NOT]
NULL.  Noted by Teodor.

17 years agoFix IS NULL and IS NOT NULL tests on row-valued expressions to conform to
Tom Lane [Thu, 28 Sep 2006 20:51:43 +0000 (20:51 +0000)]
Fix IS NULL and IS NOT NULL tests on row-valued expressions to conform to
the SQL spec, viz IS NULL is true if all the row's fields are null, IS NOT
NULL is true if all the row's fields are not null.  The former coding got
this right for a limited number of cases with IS NULL (ie, those where it
could disassemble a ROW constructor at parse time), but was entirely wrong
for IS NOT NULL.  Per report from Teodor.

I desisted from changing the behavior for arrays, since on closer inspection
it's not clear that there's any support for that in the SQL spec.  This
probably needs more consideration.

17 years agoAdd URL for client_encoding being set by client:
Bruce Momjian [Thu, 28 Sep 2006 17:43:36 +0000 (17:43 +0000)]
Add URL for client_encoding being set by client:

>   http://archives.postgresql.org/pgsql-hackers/2006-08/msg01696.php

17 years agoFix a couple other places that need to know about (some) src/port/ files
Tom Lane [Thu, 28 Sep 2006 14:48:45 +0000 (14:48 +0000)]
Fix a couple other places that need to know about (some) src/port/ files
explicitly.

17 years agoOops, forgot that libpq needs extra steps to make use of src/port/ files.
Tom Lane [Wed, 27 Sep 2006 21:29:17 +0000 (21:29 +0000)]
Oops, forgot that libpq needs extra steps to make use of src/port/ files.

17 years agoReplace strncpy with strlcpy in selected places that seem possibly relevant
Tom Lane [Wed, 27 Sep 2006 18:40:10 +0000 (18:40 +0000)]
Replace strncpy with strlcpy in selected places that seem possibly relevant
to performance.  (A wholesale effort to get rid of strncpy should be
undertaken sometime, but not during beta.)  This commit also fixes dynahash.c
to correctly truncate overlength string keys for hashtables, so that its
callers don't have to anymore.

17 years agoAdd strlcpy() to the set of functions supported by src/port/ when not
Tom Lane [Wed, 27 Sep 2006 16:29:46 +0000 (16:29 +0000)]
Add strlcpy() to the set of functions supported by src/port/ when not
available directly on the platform.  Per discussion, this function is
sufficiently widely recognized to be treated as standard.

17 years agoFix some misuses of strncat().
Tom Lane [Wed, 27 Sep 2006 16:19:42 +0000 (16:19 +0000)]
Fix some misuses of strncat().

17 years agoRemove separate strdup.h header file; it's redundant with port.h.
Tom Lane [Wed, 27 Sep 2006 15:41:24 +0000 (15:41 +0000)]
Remove separate strdup.h header file; it's redundant with port.h.

17 years agoGiST stands for "Generalized Search Tree", and therefore should not be
Neil Conway [Wed, 27 Sep 2006 06:37:28 +0000 (06:37 +0000)]
GiST stands for "Generalized Search Tree", and therefore should not be
spelled with a capital "I".

17 years agoDegrade the transaction-id wraparound point message from LOG to DEBUG1, per
Alvaro Herrera [Tue, 26 Sep 2006 17:21:39 +0000 (17:21 +0000)]
Degrade the transaction-id wraparound point message from LOG to DEBUG1, per
discussion.

Patch from Simon Riggs.

17 years agoUpdate wording of inheritance major release item.
Bruce Momjian [Tue, 26 Sep 2006 12:31:48 +0000 (12:31 +0000)]
Update wording of inheritance major release item.

17 years agoSplit out inheritance major feature item in release notes.
Bruce Momjian [Tue, 26 Sep 2006 11:45:33 +0000 (11:45 +0000)]
Split out inheritance major feature item in release notes.

17 years agoDone:
Bruce Momjian [Tue, 26 Sep 2006 11:18:29 +0000 (11:18 +0000)]
Done:

<    o Add command to archive partially filled write-ahead logs? [pitr]
<
<      Currently only full WAL files are archived. This means that the
<      most recent transactions aren't available for recovery in case
<      of a disk failure.
<

Reorder:

<    o %Create dump tool for write-ahead logs for use in determining
<      transaction id for point-in-time recovery
>    o %Create dump tool for write-ahead logs for use in determining
>      transaction id for point-in-time recovery

17 years agoAdded new files
Michael Meskes [Tue, 26 Sep 2006 09:07:31 +0000 (09:07 +0000)]
Added new files

17 years agoAdded Joachim's changes for MinGW.
Michael Meskes [Tue, 26 Sep 2006 07:56:57 +0000 (07:56 +0000)]
Added Joachim's changes for MinGW.
Added SET DATESTYLE to one test so the output format is defined.

17 years agoIncrease default value of effective_cache_size to 128MB, per discussion.
Tom Lane [Mon, 25 Sep 2006 22:12:24 +0000 (22:12 +0000)]
Increase default value of effective_cache_size to 128MB, per discussion.

17 years agoAdd a check to prevent overwriting valid data if smgrnblocks() gives a
Tom Lane [Mon, 25 Sep 2006 22:01:10 +0000 (22:01 +0000)]
Add a check to prevent overwriting valid data if smgrnblocks() gives a
wrong answer, as has been seen to occur with a buggy Linux kernel.  Not
really our bug, but it's a simple test in a seldom-used control path,
so might as well have a defense.

17 years agoMore incremental improvements for the release notes.
Neil Conway [Mon, 25 Sep 2006 15:35:28 +0000 (15:35 +0000)]
More incremental improvements for the release notes.

17 years agono-data-for-failed-tables is a pg_restore option, not a pg_dump option.
Tom Lane [Mon, 25 Sep 2006 15:31:50 +0000 (15:31 +0000)]
no-data-for-failed-tables is a pg_restore option, not a pg_dump option.

17 years agoFix notice message from DROP FUNCTION IF EXISTS, and improve message
Tom Lane [Mon, 25 Sep 2006 15:17:34 +0000 (15:17 +0000)]
Fix notice message from DROP FUNCTION IF EXISTS, and improve message
for DROP AGGREGATE IF EXISTS.  Per report from Teodor.

17 years agoFix name, "Laurenz Albe" at request of author.
Bruce Momjian [Mon, 25 Sep 2006 15:04:37 +0000 (15:04 +0000)]
Fix name,  "Laurenz Albe" at request of author.

17 years agoMinor incremental improvements to the release notes.
Neil Conway [Sun, 24 Sep 2006 18:47:56 +0000 (18:47 +0000)]
Minor incremental improvements to the release notes.

17 years agoFix incorrect mapping of fopen mode 'a' in recently-added code to
Tom Lane [Sun, 24 Sep 2006 17:19:53 +0000 (17:19 +0000)]
Fix incorrect mapping of fopen mode 'a' in recently-added code to
make fopen work safely on Windows.  Magnus

17 years agoCause pg_regress to invoke the temporary postmaster as 'postgres' not
Tom Lane [Sun, 24 Sep 2006 17:10:18 +0000 (17:10 +0000)]
Cause pg_regress to invoke the temporary postmaster as 'postgres' not
'postmaster', so as not to depend on the existence of the postmaster
symlink.  Also, implement postmaster-still-alive and postmaster-kill
operations for Windows, per Magnus.

17 years agoSuppress useless warning on pre-XP versions of Windows. Magnus
Tom Lane [Sun, 24 Sep 2006 16:59:45 +0000 (16:59 +0000)]
Suppress useless warning on pre-XP versions of Windows.  Magnus

17 years agoTag us Beta1 REL8_2_BETA1
PostgreSQL Daemon [Sat, 23 Sep 2006 01:33:37 +0000 (01:33 +0000)]
Tag us Beta1

17 years agoFix pg_locks view to call advisory locks advisory locks, while preserving
Tom Lane [Fri, 22 Sep 2006 23:20:14 +0000 (23:20 +0000)]
Fix pg_locks view to call advisory locks advisory locks, while preserving
backward compatibility for anyone using the old userlock code that's now
on pgfoundry --- locks from that code still show as 'userlock'.

17 years agoFix bugs in plpgsql and ecpg caused by assuming that isspace() would only
Tom Lane [Fri, 22 Sep 2006 21:39:58 +0000 (21:39 +0000)]
Fix bugs in plpgsql and ecpg caused by assuming that isspace() would only
return true for exactly the characters treated as whitespace by their flex
scanners.  Per report from Victor Snezhko and subsequent investigation.

Also fix a passel of unsafe usages of <ctype.h> functions, that is, ye olde
char-vs-unsigned-char issue.  I won't miss <ctype.h> when we are finally
able to stop using it.

17 years agoSurely this temp buffer needn't be static.
Tom Lane [Fri, 22 Sep 2006 19:51:14 +0000 (19:51 +0000)]
Surely this temp buffer needn't be static.

17 years agoRearrange yes/no prompting code so that the prompts always show the
Peter Eisentraut [Fri, 22 Sep 2006 18:50:41 +0000 (18:50 +0000)]
Rearrange yes/no prompting code so that the prompts always show the
(possibly (un)translated) letters that are actually expected as input.
Also reject invalid responses instead of silenty taken them as "no".

with help from Bernd Helmle

17 years agoAdd to major release notes items, per Simon.
Bruce Momjian [Fri, 22 Sep 2006 18:09:40 +0000 (18:09 +0000)]
Add to major release notes items, per Simon.

17 years agoAdd units to the default postgresql.conf. For the most part, this should
Peter Eisentraut [Fri, 22 Sep 2006 17:41:21 +0000 (17:41 +0000)]
Add units to the default postgresql.conf.  For the most part, this should
match what SHOW displays as default value, to make the user experience
uniform.

17 years agoMove incorrectly-located indexterm entry.
Tom Lane [Fri, 22 Sep 2006 16:35:55 +0000 (16:35 +0000)]
Move incorrectly-located indexterm entry.

17 years agoUpdate timezone documentation to reflect current reality: instead of
Tom Lane [Fri, 22 Sep 2006 16:20:00 +0000 (16:20 +0000)]
Update timezone documentation to reflect current reality: instead of
giving tables of known timezone names, refer the user to the system views.
Joachim Wieland