OSDN Git Service

pg-rex/syncrep.git
20 years agoXLOG file archiving and point-in-time recovery. There are still some
Tom Lane [Mon, 19 Jul 2004 02:47:16 +0000 (02:47 +0000)]
XLOG file archiving and point-in-time recovery.  There are still some
loose ends and a glaring lack of documentation, but it basically works.

Simon Riggs with some editorialization by Tom Lane.

20 years agoRename AC_PROG_LD* macros to PGAC_PROG_LD*. This avoids clashes with the
Peter Eisentraut [Sat, 17 Jul 2004 18:53:56 +0000 (18:53 +0000)]
Rename AC_PROG_LD* macros to PGAC_PROG_LD*.  This avoids clashes with the
macros provided by the real libtool, when other packages borrow some macros
from PostgreSQL, as in the case of the ODBC driver.

20 years agoWhen renaming a column that participates in a foreign key, we must
Tom Lane [Sat, 17 Jul 2004 17:28:29 +0000 (17:28 +0000)]
When renaming a column that participates in a foreign key, we must
force relcache rebuild for the other table as well as the column's
own table.  Otherwise, already-cached foreign key triggers will stop
working.  Per example from Alexander Pravking.

20 years agoFix broken markup, per Jonathan Gardner.
Tom Lane [Sat, 17 Jul 2004 16:33:31 +0000 (16:33 +0000)]
Fix broken markup, per Jonathan Gardner.

20 years agoAdd:
Bruce Momjian [Sat, 17 Jul 2004 10:50:05 +0000 (10:50 +0000)]
Add:

>  o Enforce rules for setting combinations

20 years agoInvent ResourceOwner mechanism as per my recent proposal, and use it to
Tom Lane [Sat, 17 Jul 2004 03:32:14 +0000 (03:32 +0000)]
Invent ResourceOwner mechanism as per my recent proposal, and use it to
keep track of portal-related resources separately from transaction-related
resources.  This allows cursors to work in a somewhat sane fashion with
nested transactions.  For now, cursor behavior is non-subtransactional,
that is a cursor's state does not roll back if you abort a subtransaction
that fetched from the cursor.  We might want to change that later.

20 years agoHere is a patch required to build plperl with win32. The issues were:
Bruce Momjian [Fri, 16 Jul 2004 19:18:24 +0000 (19:18 +0000)]
Here is a patch required to build plperl with win32. The issues were:

* perl_useshrplib gets set to "yes" and not to "true". I assume it's set
to "true" on unix, so I left both.
* Need to translate backslashes into slashes
* The linker config coming out of perl was for MSVC and not for mingw

Magnus Hagander

20 years agoProperly order mysql links.
Bruce Momjian [Thu, 15 Jul 2004 18:53:44 +0000 (18:53 +0000)]
Properly order mysql links.

20 years agoAdd French FAQ.
Bruce Momjian [Thu, 15 Jul 2004 15:25:25 +0000 (15:25 +0000)]
Add French FAQ.

LELARGE Guillaume

20 years ago> pg_ctl --help
Bruce Momjian [Thu, 15 Jul 2004 15:23:32 +0000 (15:23 +0000)]
> pg_ctl --help
>  ---skiped---
>   -P       user name of account to register PostgreSQL server
>   -U       password  of account to register PostgreSQL server
> ---skiped---
>
> I think that isn't right ;)

Claudio Natoli

20 years agoFor a SQL function declared to return a named composite type, make
Tom Lane [Thu, 15 Jul 2004 13:51:38 +0000 (13:51 +0000)]
For a SQL function declared to return a named composite type, make
sure the tuple datums it returns actually show that type and not RECORD.

20 years agoAdd permission display to \db+.
Bruce Momjian [Thu, 15 Jul 2004 03:56:06 +0000 (03:56 +0000)]
Add permission display to \db+.

20 years agoDetect locale/encoding mismatch in initdb, or pick a suitable encoding
Peter Eisentraut [Wed, 14 Jul 2004 17:55:10 +0000 (17:55 +0000)]
Detect locale/encoding mismatch in initdb, or pick a suitable encoding
automatically if none was specified.

20 years agoHave \dn+ show permissions and description for schemas.
Bruce Momjian [Tue, 13 Jul 2004 16:48:16 +0000 (16:48 +0000)]
Have \dn+ show permissions and description for schemas.

Dennis Bjorklund

20 years agoChange pg_dump to use ALTER OWNER commands instead of SET SESSION
Bruce Momjian [Tue, 13 Jul 2004 03:00:17 +0000 (03:00 +0000)]
Change pg_dump to use ALTER OWNER commands instead of SET SESSION
AUTHORIZATION commands by default.  Move all GRANT and REVOKE commands
to the end of the dump to avoid restore failures in several situations.
Bring back --use-set-session-authorization option to get previous SET
behaviour

Christopher Kings-Lyne

20 years agoCan't have blank lines in resultmap; some versions of expr complain.
Tom Lane [Tue, 13 Jul 2004 02:55:25 +0000 (02:55 +0000)]
Can't have blank lines in resultmap; some versions of expr complain.

20 years agoAdd new join file.
Bruce Momjian [Tue, 13 Jul 2004 02:54:08 +0000 (02:54 +0000)]
Add new join file.

20 years agoMove tablespace name up on to the same line in psql \d.
Bruce Momjian [Tue, 13 Jul 2004 02:46:21 +0000 (02:46 +0000)]
Move tablespace name up on to the same line in psql \d.

20 years agoRename join ordering named join_1.out
Bruce Momjian [Tue, 13 Jul 2004 02:05:09 +0000 (02:05 +0000)]
Rename join ordering named join_1.out

20 years agoCause libpq and ecpg libraries to be built as proper shared libraries
Tom Lane [Tue, 13 Jul 2004 00:06:46 +0000 (00:06 +0000)]
Cause libpq and ecpg libraries to be built as proper shared libraries
(.dylib format) on Mac OS X, while not messing up loadable modules for
the backend (which are the same kind of animal as a shared library on
every other platform, but not here).  Also get the naming convention
to match OS X practice, viz libFOO.version.so not libFOO.so.version.
In support of that last, refactor code in Makefile.shlib to make it
easier to have platform-specific shlib naming conventions.
This patch is loosely based on the Fink project's current postgresql.patch.
Tested by yours truly on OS X 10.3.4; does anyone have 10.2.* to check
it on?

20 years agoMake installation instructions match reality.
Peter Eisentraut [Mon, 12 Jul 2004 22:12:34 +0000 (22:12 +0000)]
Make installation instructions match reality.

20 years agoAdd tablespace location display for psql \d.
Bruce Momjian [Mon, 12 Jul 2004 20:41:13 +0000 (20:41 +0000)]
Add tablespace location display for psql \d.

Gavin Sherry

20 years agoAdd has_tablespace_privilege().
Bruce Momjian [Mon, 12 Jul 2004 20:23:59 +0000 (20:23 +0000)]
Add has_tablespace_privilege().

Christopher Kings-Lynne

20 years agoAdd Win32-specific join ordering.
Bruce Momjian [Mon, 12 Jul 2004 20:11:37 +0000 (20:11 +0000)]
Add Win32-specific join ordering.

20 years agoAdd blank line to new canonicalize_path file, which handles "a b\" properly.
Bruce Momjian [Mon, 12 Jul 2004 19:27:31 +0000 (19:27 +0000)]
Add blank line to new canonicalize_path file, which handles "a b\" properly.

20 years agocanonicalization adjustments.
Bruce Momjian [Mon, 12 Jul 2004 19:18:18 +0000 (19:18 +0000)]
canonicalization adjustments.

20 years agoPut back canonicalization of PGDATA environment variable.
Bruce Momjian [Mon, 12 Jul 2004 19:15:14 +0000 (19:15 +0000)]
Put back canonicalization of PGDATA environment variable.

20 years agoUse strdup in pg_ctl for canonicalize_path on environment variable.
Bruce Momjian [Mon, 12 Jul 2004 18:17:13 +0000 (18:17 +0000)]
Use strdup in pg_ctl for canonicalize_path on environment variable.
Simplify postmaster call too.

20 years agoThis patch adds the following options to pg_dumpall, to be passed to
Bruce Momjian [Mon, 12 Jul 2004 14:35:45 +0000 (14:35 +0000)]
This patch adds the following options to pg_dumpall, to be passed to
pg_dump:

-S, --superuser=NAME

-O, --no-owner

-X disable-dollar-quoting, --disable-dollar-quoting

-X disable-triggers, --disable-triggers

Christopher Kings-Lynne

20 years agoThe attached patch, which incorporates the previous one sent and
Bruce Momjian [Mon, 12 Jul 2004 14:31:04 +0000 (14:31 +0000)]
The attached patch, which incorporates the previous one sent and
currently unapplied regarding spi_internal.c, makes some additional
fixes relating to return types, and also contains the fix for
preventing  the use of insecure versions of Safe.pm.

There is one remaing return case that does not appear to work, namely
return of a composite directly in a select, i.e. if  foo returns some
composite type, 'select * from foo()' works but 'select foo()' doesn't.
We will either fix that or document it as a limitation.

The function plperl_func_handler is a mess - I will try to get it
cleaned up (and split up) in a subsequent patch, time permitting.

Also, reiterating previous advice - this changes slightly the API for
spi_exec_query - the returned object has either 2 or 3 members: 'status'
(string) and 'proceesed' (int,- number of rows) and, if rows are
returned, 'rows' (array of tuple hashes).

Andrew Dunstan

20 years ago> win32 doesn't support a static initializer for mutexes, thus the first
Bruce Momjian [Mon, 12 Jul 2004 14:23:28 +0000 (14:23 +0000)]
> win32 doesn't support a static initializer for mutexes, thus the first
> user must initialize the lock. The problem are concurrent "first" users
> - the pthread_mutex_t initialization must be synchronized.
> The current implementation is broken, the attached patches fixes that:
> mutex_initlock is a spinlock. If the pthread_mutex_t mutex is not
> initialized, then the spinlock is acquired, if the pthread_mutex_t is
> initialized if it's not yet initialized and then the spinlock is dropped.

Manfred Spraul

20 years agoBack out thread fix until I get clarification.
Bruce Momjian [Mon, 12 Jul 2004 14:16:28 +0000 (14:16 +0000)]
Back out thread fix until I get clarification.

20 years agowin32 doesn't support a static initializer for mutexes, thus the first
Bruce Momjian [Mon, 12 Jul 2004 14:11:17 +0000 (14:11 +0000)]
win32 doesn't support a static initializer for mutexes, thus the first
user must initialize the lock. The problem are concurrent "first" users
- the pthread_mutex_t initialization must be synchronized.
The current implementation is broken, the attached patches fixes that:
mutex_initlock is a spinlock. If the pthread_mutex_t mutex is not
initialized, then the spinlock is acquired, if the pthread_mutex_t is
initialized if it's not yet initialized and then the spinlock is
dropped.

Manfred Spraul

20 years agoRemove TABLESPACE option of CREATE SEQUENCE; sequences will now always
Tom Lane [Mon, 12 Jul 2004 05:38:11 +0000 (05:38 +0000)]
Remove TABLESPACE option of CREATE SEQUENCE; sequences will now always
live in database or schema's default tablespace, as per today's discussion.
Also, remove some unused keywords from the grammar (PATH, PENDANT,
VERSION), and fix ALSO, which was added as a keyword but not added
to the keyword classification lists, thus making it worse-than-reserved.

20 years agoNew text.
Bruce Momjian [Mon, 12 Jul 2004 03:58:34 +0000 (03:58 +0000)]
New text.

20 years agoFix library_path with canonicalization.
Bruce Momjian [Mon, 12 Jul 2004 02:22:51 +0000 (02:22 +0000)]
Fix library_path with canonicalization.

20 years agoFix username mismatch in initdb. Magnus.
Bruce Momjian [Mon, 12 Jul 2004 01:54:10 +0000 (01:54 +0000)]
Fix username mismatch in initdb.  Magnus.

20 years agoWording improvement.
Bruce Momjian [Mon, 12 Jul 2004 01:37:01 +0000 (01:37 +0000)]
Wording improvement.

20 years agoTablespace examples for CREATE TABLE/INDEX/SCHEMA/DATABASE as well as
Bruce Momjian [Mon, 12 Jul 2004 01:22:53 +0000 (01:22 +0000)]
Tablespace examples for CREATE TABLE/INDEX/SCHEMA/DATABASE as well as
some other examples for CREATE DATABASE.

Gavin Sherry

20 years agoMove comma.
Bruce Momjian [Mon, 12 Jul 2004 01:10:17 +0000 (01:10 +0000)]
Move comma.

20 years agoThe outer #define was forgotten. Attached patch adds it; please apply.
Bruce Momjian [Mon, 12 Jul 2004 01:00:45 +0000 (01:00 +0000)]
The outer #define was forgotten.  Attached patch adds it; please apply.

Alvaro

20 years agoAdd:
Bruce Momjian [Mon, 12 Jul 2004 00:39:11 +0000 (00:39 +0000)]
Add:

> * Add session start time and last statement time to pg_stat_activity

20 years agoAdded Matthew:
Bruce Momjian [Mon, 12 Jul 2004 00:23:59 +0000 (00:23 +0000)]
Added Matthew:

<   rather than in /contrib
>   rather than in /contrib (Matthew)
515a516
> * Matthew T. O'Connor <matthew@zeut.net>

20 years agoNew Czech FAQ.
Bruce Momjian [Mon, 12 Jul 2004 00:22:51 +0000 (00:22 +0000)]
New Czech FAQ.

Pavel Stehule

20 years agoAdd FAQ entry on hardware selection.
Bruce Momjian [Mon, 12 Jul 2004 00:18:57 +0000 (00:18 +0000)]
Add FAQ entry on hardware selection.

20 years agoCanonicalize preload_libraries after it is split up, not before.
Bruce Momjian [Mon, 12 Jul 2004 00:09:07 +0000 (00:09 +0000)]
Canonicalize preload_libraries after it is split up, not before.

20 years agoRemove:
Bruce Momjian [Sun, 11 Jul 2004 23:52:05 +0000 (23:52 +0000)]
Remove:

< * Allow moving sequences and toast tables to other tablespaces

20 years agoCleanup for canonicalization fixes, from Tom.
Bruce Momjian [Sun, 11 Jul 2004 23:49:51 +0000 (23:49 +0000)]
Cleanup for canonicalization fixes, from Tom.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 23:31:27 +0000 (23:31 +0000)]
Add:

> * Allow moving sequences and toast tables to other tablespaces

20 years agoThis patch fixes a small error in the Porting PL/SQL to PL/pgSQL
Bruce Momjian [Sun, 11 Jul 2004 23:26:51 +0000 (23:26 +0000)]
This patch fixes a small error in the Porting PL/SQL to PL/pgSQL
section where a instr function parameter is mistyped as varchar. It
works properly when changed to integer.

Michael Glaesemann

20 years agoExample for create function using argument names
Bruce Momjian [Sun, 11 Jul 2004 23:23:43 +0000 (23:23 +0000)]
Example for create function using argument names

Gavin Sherry

20 years agoALTER TABLE SET TABLESPACE. Gavin Sherry, some rework by Tom Lane.
Tom Lane [Sun, 11 Jul 2004 23:13:58 +0000 (23:13 +0000)]
ALTER TABLE SET TABLESPACE.  Gavin Sherry, some rework by Tom Lane.

20 years agoUpdate German FAQ>
Bruce Momjian [Sun, 11 Jul 2004 23:07:18 +0000 (23:07 +0000)]
Update German FAQ>

20 years agoUpdate URL's for mysql conversion tools.
Bruce Momjian [Sun, 11 Jul 2004 22:52:18 +0000 (22:52 +0000)]
Update URL's for mysql conversion tools.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 21:57:03 +0000 (21:57 +0000)]
Add:

> * Add NOVICE output level for helpful messages like automatic sequence/index creation

20 years agoRemove postgresql.conf of 'info' as a valid client_min_messages level.
Bruce Momjian [Sun, 11 Jul 2004 21:48:25 +0000 (21:48 +0000)]
Remove postgresql.conf of 'info' as a valid client_min_messages level.

20 years agoUse canonicalize_path for -D, GUC paths, and paths coming in from
Bruce Momjian [Sun, 11 Jul 2004 21:34:04 +0000 (21:34 +0000)]
Use canonicalize_path for -D, GUC paths, and paths coming in from
environment variables.

20 years agoMove TablespaceCreateDbspace() call into smgrcreate(), which is where it
Tom Lane [Sun, 11 Jul 2004 19:52:52 +0000 (19:52 +0000)]
Move TablespaceCreateDbspace() call into smgrcreate(), which is where it
probably should have been to begin with; this is to cover cases like
needing to recreate the per-db directory during WAL replay.
Also, fix heap_create to force pg_class.reltablespace to be zero instead
of the database's default tablespace; this makes the world safe for
CREATE DATABASE to handle all tables in the default tablespace alike,
as per previous discussion.  And force pg_class.reltablespace to zero
when creating a relation without physical storage (eg, a view); this
avoids possibly having dangling references in this column after a
subsequent DROP TABLESPACE.

20 years agoRename XLOG_BTREE_NEWPAGE xlog record type into XLOG_HEAP_NEWPAGE, and
Tom Lane [Sun, 11 Jul 2004 18:01:45 +0000 (18:01 +0000)]
Rename XLOG_BTREE_NEWPAGE xlog record type into XLOG_HEAP_NEWPAGE, and
shift support code into heapam.c accordingly.  This is in service of
soon-to-be-committed ALTER TABLE SET TABLESPACE code that will want to
use this same record type for both heaps and indexes.

Theoretically I should have forced initdb for this, but in practice there
is no change in xlog contents because CVS tip will never really emit this
record type anyhow...

20 years agoUse standard macro for psql binary file open. Add comment explaining
Bruce Momjian [Sun, 11 Jul 2004 13:29:16 +0000 (13:29 +0000)]
Use standard macro for psql binary file open.  Add comment explaining
control-z requirement.

20 years agoCause the format of BC timestamptz output to be 'datetime zone BC' rather
Tom Lane [Sun, 11 Jul 2004 04:57:20 +0000 (04:57 +0000)]
Cause the format of BC timestamptz output to be 'datetime zone BC' rather
than 'datetime BC zone', because the former is accepted by the timestamptz
input converter while the latter may not be depending on spacing.  This
is not a loss of compatibility w.r.t. 7.4 and before, because until very
recently there was never a case where we'd output both zone and 'BC'.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 03:10:50 +0000 (03:10 +0000)]
Add:

> * Allow moving system tables to other tablespaces, where possible

20 years agoFix trim_trailing_separator() to not trim c:\ nor \\network\ on Win32.
Bruce Momjian [Sun, 11 Jul 2004 02:59:42 +0000 (02:59 +0000)]
Fix trim_trailing_separator() to not trim c:\ nor \\network\ on Win32.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 02:03:22 +0000 (02:03 +0000)]
Add:

> * Allow changing of already-created database and schema tablespaces

20 years agoOpen files in binary mode on Win32 so control-z isn't seen as EOF.
Bruce Momjian [Sun, 11 Jul 2004 00:54:55 +0000 (00:54 +0000)]
Open files in binary mode on Win32 so control-z isn't seen as EOF.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 00:28:30 +0000 (00:28 +0000)]
Add:

> *  Add an option to sync() before fsync()'ing checkpoint files

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 00:26:36 +0000 (00:26 +0000)]
Add:

<
> * Use a phantom command counter for nested subtransactions to reduce
>   tuple overhead

20 years agoAllow configuration files to be placed outside the data directory.
Bruce Momjian [Sun, 11 Jul 2004 00:18:45 +0000 (00:18 +0000)]
Allow configuration files to be placed outside the data directory.

Add new postgresql.conf variables to point to data, pg_hba.conf, and
pg_ident.conf files.

Needs more documentation.

20 years agoAdd:
Bruce Momjian [Sun, 11 Jul 2004 00:18:28 +0000 (00:18 +0000)]
Add:

> * Add include functionality to postgresql.conf
>

20 years agoDefend against overrun of ExtraOptions array --- strictly paranoia,
Tom Lane [Sat, 10 Jul 2004 23:29:16 +0000 (23:29 +0000)]
Defend against overrun of ExtraOptions array --- strictly paranoia,
since the person or script starting the postmaster has to be trusted
anyway.

20 years agoCheck more test points (in fact, every week in 1970..2004) to get a more
Tom Lane [Sat, 10 Jul 2004 23:06:50 +0000 (23:06 +0000)]
Check more test points (in fact, every week in 1970..2004) to get a more
accurate matching of our time zone to the system's zone.  This method is
able to distinguish Antarctica/Casey from Australia/Perth, as in Chris
K-L's recent example; and it is not materially slower than before, because
the extra checks generally don't get done against very many time zones.

It seems possible that with this test we'd be able to correctly identify
Windows timezones without looking at the timezone name, but I do not
have the ability to try it.

20 years agoFix BSD-only coding in port.c (passing a local variable to putenv).
Tom Lane [Sat, 10 Jul 2004 22:58:42 +0000 (22:58 +0000)]
Fix BSD-only coding in port.c (passing a local variable to putenv).
Also a quick but half-baked attempt to make trim_trailing_separator
do the right thing with path consisting only of '/' --- still needs
work for Windows I think.

20 years agoTest HAVING condition before computing targetlist of an Aggregate node.
Tom Lane [Sat, 10 Jul 2004 18:39:23 +0000 (18:39 +0000)]
Test HAVING condition before computing targetlist of an Aggregate node.
This is required by SQL spec to avoid failures in cases like
  SELECT sum(win)/sum(lose) FROM ... GROUP BY ... HAVING sum(lose) > 0;
AFAICT we have gotten this wrong since day one.  Kudos to Holger Jakobs
for being the first to notice.

20 years agoClarify schema meanings in pg_dump manual page by referring to object
Bruce Momjian [Sat, 10 Jul 2004 15:51:28 +0000 (15:51 +0000)]
Clarify schema meanings in pg_dump manual page by referring to object
definitions more frequently.

20 years agoAdd --enable-thread-safety-force to force a thread compile even if the
Bruce Momjian [Sat, 10 Jul 2004 01:24:29 +0000 (01:24 +0000)]
Add --enable-thread-safety-force to force a thread compile even if the
OS thread tests fail.

20 years agoAdd OSX thread support for next 7.4.X
Bruce Momjian [Wed, 7 Jul 2004 13:04:45 +0000 (13:04 +0000)]
Add OSX thread support for next 7.4.X

20 years agoFormatting improvements:
Bruce Momjian [Wed, 7 Jul 2004 12:48:54 +0000 (12:48 +0000)]
Formatting improvements:

< The most recent version of this document can be viewed at
< the PostgreSQL web site, http://www.PostgreSQL.org.
> The most recent version of this document can be viewed at the PostgreSQL web site, http://www.PostgreSQL.org.

20 years agoRegenerate:
Bruce Momjian [Wed, 7 Jul 2004 12:45:50 +0000 (12:45 +0000)]
Regenerate:

> #A dash (-) marks changes that will appear in the upcoming 7.5 release.#

20 years agoregenerate.
Bruce Momjian [Wed, 7 Jul 2004 12:45:16 +0000 (12:45 +0000)]
regenerate.

20 years agoBold dash line.
Bruce Momjian [Wed, 7 Jul 2004 12:44:34 +0000 (12:44 +0000)]
Bold dash line.
> \#A dash (-) marks changes that will appear in the upcoming 7.5 release.#

20 years agoUpdate highlighting of dash.
Bruce Momjian [Wed, 7 Jul 2004 12:25:02 +0000 (12:25 +0000)]
Update highlighting of dash.

20 years agoMove section:
Bruce Momjian [Tue, 6 Jul 2004 20:29:31 +0000 (20:29 +0000)]
Move section:

< * Add replication of distributed databases
<  o Automatic failover
<  o Load balancing
<  o Master/slave replication
<  o Multi-master replication
<  o Partition data across servers
<  o Queries across databases or servers (two-phase commit)
<  o Allow replication over unreliable or non-persistent links
55a48,55
> * Improve replication solutions
>  o Automatic failover
>  o Load balancing
>  o Master/slave replication
>  o Multi-master replication
>  o Partition data across servers
>  o Queries across databases or servers (two-phase commit)
>  o Allow replication over unreliable or non-persistent links

20 years agoAdd comment describing ereport() NOTICE/WARNING distinction.
Bruce Momjian [Tue, 6 Jul 2004 19:51:59 +0000 (19:51 +0000)]
Add comment describing ereport() NOTICE/WARNING distinction.

20 years agoReorder dash mention to be more prominient.
Bruce Momjian [Tue, 6 Jul 2004 16:18:31 +0000 (16:18 +0000)]
Reorder dash mention to be more prominient.

> A dash (-) marks changes that will appear in the upcoming 7.5 release.
>
> Bracketed items "[]" have more detail.
9,12d12
<
< A dash (-) marks changes that will appear in the upcoming 7.5 release.
<
< Bracketed items "[]" have more detail.

20 years agoFix broken logic for pretty-printing parenthesis-suppression in UNION
Tom Lane [Tue, 6 Jul 2004 04:50:21 +0000 (04:50 +0000)]
Fix broken logic for pretty-printing parenthesis-suppression in UNION
et al.

20 years agoDone:
Bruce Momjian [Tue, 6 Jul 2004 03:23:31 +0000 (03:23 +0000)]
Done:

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

20 years agoRevert broken rpath addition.
Tom Lane [Mon, 5 Jul 2004 23:24:12 +0000 (23:24 +0000)]
Revert broken rpath addition.

20 years agoFix unchecked mallocs/strdups added by recent placeholder-config-vars
Tom Lane [Mon, 5 Jul 2004 23:14:14 +0000 (23:14 +0000)]
Fix unchecked mallocs/strdups added by recent placeholder-config-vars
patch.

Thomas Hallgren

20 years agoAdded free() calls against memory leak in interval.c.
Michael Meskes [Mon, 5 Jul 2004 18:51:51 +0000 (18:51 +0000)]
Added free() calls against memory leak in interval.c.

20 years ago- Fixed indicator in SET DESCRIPTOR.
Michael Meskes [Mon, 5 Jul 2004 09:45:54 +0000 (09:45 +0000)]
- Fixed indicator in SET DESCRIPTOR.
- Added special handling of descriptor header information.
- Some code cleanup.

20 years agoFix paragraph mention in ALTER DATABASE manual page.
Bruce Momjian [Mon, 5 Jul 2004 02:34:43 +0000 (02:34 +0000)]
Fix paragraph mention in ALTER DATABASE manual page.

Christopher Kings-Lynne

20 years agoAllow plperl to compile with Perl 5.05.
Bruce Momjian [Mon, 5 Jul 2004 02:18:31 +0000 (02:18 +0000)]
Allow plperl to compile with Perl 5.05.

Andrew Dunstan

20 years agoFix misspellings: langauge -> language.
Tom Lane [Sun, 4 Jul 2004 23:34:24 +0000 (23:34 +0000)]
Fix misspellings: langauge -> language.

20 years agoMade sure SET DESCRIPTOR accepts all data types including constants.
Michael Meskes [Sun, 4 Jul 2004 15:02:24 +0000 (15:02 +0000)]
Made sure SET DESCRIPTOR accepts all data types including constants.

20 years agoOkay, I've had it with answering newbie questions about why plpgsql
Tom Lane [Sun, 4 Jul 2004 02:49:04 +0000 (02:49 +0000)]
Okay, I've had it with answering newbie questions about why plpgsql
FOR loops are giving weird syntax errors.  Restructure parsing of FOR
loops so that the integer-loop-vs-query-loop decision is driven off
the presence of '..' between IN and LOOP, rather than the presence
of a matching record/row variable name.  Hopefully this will make the
behavior a bit more transparent.

20 years agoFix no-longer-correct bit-pushing in TransactionIdSetStatus, per Alvaro.
Tom Lane [Sat, 3 Jul 2004 02:55:56 +0000 (02:55 +0000)]
Fix no-longer-correct bit-pushing in TransactionIdSetStatus, per Alvaro.

20 years agoAdd missing operators of the form interval-plus-datetime, as required for
Tom Lane [Fri, 2 Jul 2004 22:50:23 +0000 (22:50 +0000)]
Add missing operators of the form interval-plus-datetime, as required for
better SQL compliance in this area, per recent discussion.  Mark related
operators as commutators where possible.  (The system doesn't actually care
about commutator marking for operators not returning boolean, at the moment,
but this seems forward-thinking and besides it made it easier to verify
that we hadn't missed any.)
Also, remove interval-minus-time and interval-minus-timetz operators.
I'm not sure how these got in, but they are nonstandard and had very
obviously broken behavior.  (minus is not commutative in anyone's book.)
I doubt anyone had ever used 'em, because we'd surely have gotten a bug
report about it if so.

20 years agoAndreas Pflug wrote:
Joe Conway [Fri, 2 Jul 2004 18:59:25 +0000 (18:59 +0000)]
Andreas Pflug wrote:
 From an idea of Bruce, the attached patch implements the function
 pg_tablespace_databases(oid) RETURNS SETOF oid
 which delivers as set of database oids having objects in the selected
 tablespace, enabling an admin to examine only the databases affecting
 the tablespace for objects instead of scanning all of them.

initdb forced

20 years agoTranslation update
Peter Eisentraut [Fri, 2 Jul 2004 15:07:10 +0000 (15:07 +0000)]
Translation update

20 years agoAttached is a 1 line bug fix for dbmirror that was submitted.
Joe Conway [Fri, 2 Jul 2004 00:58:09 +0000 (00:58 +0000)]
Attached is a 1 line bug fix for dbmirror that was submitted.
It fixes a bug where some transactions could be dropped when writing
mirrored SQL statements to files.
-- Steven Singer