]> granicus.if.org Git - postgresql/log
postgresql
17 years agoRemove tabs from SGML reference files, so their addition can be detected
Bruce Momjian [Wed, 21 Feb 2007 23:21:12 +0000 (23:21 +0000)]
Remove tabs from SGML reference files, so their addition can be detected
in the future.

17 years agoUpdate upgrade wording.
Bruce Momjian [Wed, 21 Feb 2007 22:54:11 +0000 (22:54 +0000)]
Update upgrade wording.

17 years agoUpdate new optional VACUUM FULL hint for translations, per Alvaro.
Bruce Momjian [Wed, 21 Feb 2007 22:47:45 +0000 (22:47 +0000)]
Update new optional VACUUM FULL hint for translations, per Alvaro.

17 years agoMove increase FSM warning to after lazy_truncate_heap() because the
Bruce Momjian [Wed, 21 Feb 2007 22:15:21 +0000 (22:15 +0000)]
Move increase FSM warning to after lazy_truncate_heap() because the
function might reduce the number of free pages in the table.  Recommend
VACUUM FULL only if 20% free.

Simon Riggs.

17 years agobtree source code cleanups:
Bruce Momjian [Wed, 21 Feb 2007 20:02:17 +0000 (20:02 +0000)]
btree source code cleanups:

I refactored findsplitloc and checksplitloc so that the division of
labor is more clear IMO. I pushed all the space calculation inside the
loop to checksplitloc.

I also fixed the off by 4 in free space calculation caused by
PageGetFreeSpace subtracting sizeof(ItemIdData), even though it was
harmless, because it was distracting and I felt it might come back to
bite us in the future if we change the page layout or alignments.
There's now a new function PageGetExactFreeSpace that doesn't do the
subtraction.

findsplitloc now tries the "just the new item to right page" split as
well. If people don't like the refactoring, I can write a patch to just
add that.

Heikki Linnakangas

17 years agoUpdate FAQ about minor updates.
Bruce Momjian [Wed, 21 Feb 2007 19:48:02 +0000 (19:48 +0000)]
Update FAQ about minor updates.

17 years agoUpdate Japanese FAQ.
Bruce Momjian [Wed, 21 Feb 2007 17:23:49 +0000 (17:23 +0000)]
Update Japanese FAQ.

Jun Kuwamura

17 years agoUpdate URL in German FAQ for bug form, per Schima, Fabian
Bruce Momjian [Wed, 21 Feb 2007 16:49:37 +0000 (16:49 +0000)]
Update URL in German FAQ for bug form, per Schima, Fabian

17 years agoUpdate Chinese FAQs to have two versions, a traditional Chinese version (Taiwan)
Bruce Momjian [Wed, 21 Feb 2007 16:42:36 +0000 (16:42 +0000)]
Update Chinese FAQs to have two versions, a traditional Chinese version (Taiwan)
and a Simplified version (China (PRC)).

Backpatch to 8.2.X.

Daojing.Zhou

17 years agoFix typo, per Dave Page.
Bruce Momjian [Wed, 21 Feb 2007 16:09:01 +0000 (16:09 +0000)]
Fix typo, per Dave Page.

17 years agoUpdate minor release text.
Bruce Momjian [Wed, 21 Feb 2007 16:08:05 +0000 (16:08 +0000)]
Update minor release text.

17 years agoAdd configure --enable-profiling to enable GCC profiling. Patches from
Bruce Momjian [Wed, 21 Feb 2007 15:12:39 +0000 (15:12 +0000)]
Add configure --enable-profiling to enable GCC profiling.  Patches from
Korry Douglas and Nikhil S

17 years agoUpdate upgrade text.
Bruce Momjian [Wed, 21 Feb 2007 15:07:35 +0000 (15:07 +0000)]
Update upgrade text.

17 years agoRevert changes to process pg_proc.h entries without OIDs. We're not supposed
Magnus Hagander [Wed, 21 Feb 2007 12:28:27 +0000 (12:28 +0000)]
Revert changes to process pg_proc.h entries without OIDs. We're not supposed
to have such entries, and want to be notified when we do...
Leave the plain bugfix in genbki.

17 years agoUpdate URL for set-returning functions.
Bruce Momjian [Wed, 21 Feb 2007 03:50:57 +0000 (03:50 +0000)]
Update URL for set-returning functions.

17 years agoAllow pltcl args to spi_prepare and plpython args to plpy.prepare to be standard...
Andrew Dunstan [Wed, 21 Feb 2007 03:27:32 +0000 (03:27 +0000)]
Allow pltcl args to spi_prepare and plpython args to plpy.prepare to be standard type aliases as well as those known in pg_type. Similar to recent change in plperl.

17 years agoAdd URL for:
Bruce Momjian [Wed, 21 Feb 2007 01:08:58 +0000 (01:08 +0000)]
Add URL for:

      o Allow row and record variables to be set to NULL constants,
        and allow NULL tests on such variables

17 years agoPrevent BLCKSZ < 1024, and have initdb test shared buffers based on the
Bruce Momjian [Tue, 20 Feb 2007 23:49:38 +0000 (23:49 +0000)]
Prevent BLCKSZ < 1024, and have initdb test shared buffers based on the
BLCKSZ value.

17 years agoAdd:
Bruce Momjian [Tue, 20 Feb 2007 23:39:03 +0000 (23:39 +0000)]
Add:

>
> * Consider decreasing the amount of memory used by PrivateRefCount
>
>   http://archives.postgresql.org/pgsql-hackers/2006-11/msg00797.php
>   http://archives.postgresql.org/pgsql-hackers/2007-01/msg00752.php
>

17 years agoFix markup.
Bruce Momjian [Tue, 20 Feb 2007 23:14:19 +0000 (23:14 +0000)]
Fix markup.

17 years agoChange $(CC) to $(COMPILER) on Solaris gcc so -m64 is passed into the
Bruce Momjian [Tue, 20 Feb 2007 22:45:57 +0000 (22:45 +0000)]
Change $(CC) to $(COMPILER) on Solaris gcc so -m64 is passed into the
shared link line.

17 years agoAdd:
Bruce Momjian [Tue, 20 Feb 2007 22:29:28 +0000 (22:29 +0000)]
Add:

> * Increase locking when DROPing objects so dependent objects cannot
>   get dropped while the DROP operation is happening
>
>   http://archives.postgresql.org/pgsql-hackers/2007-01/msg00937.php

17 years agoAdd URL for:
Bruce Momjian [Tue, 20 Feb 2007 22:23:05 +0000 (22:23 +0000)]
Add URL for:

        o Allow UPDATE/DELETE WHERE CURRENT OF cursor
>    http://archives.postgresql.org/pgsql-hackers/2007-01/msg01014.php

17 years agoAdd:
Bruce Momjian [Tue, 20 Feb 2007 21:22:35 +0000 (21:22 +0000)]
Add:

>
> * Add missing operators for geometric data types
>
>   Some geometric types do not have the full suite of geometric operators,
>   e.g. box @> point
>

17 years agoUpdate "encode" documentation to mention that 'escape' only changes null
Bruce Momjian [Tue, 20 Feb 2007 19:59:04 +0000 (19:59 +0000)]
Update "encode" documentation to mention that 'escape' only changes null
bytes and backslashes, remove "ASCII" mention.  Backpatch to 8.2.X.

17 years agoUpdate pgpass Win32 wording.
Bruce Momjian [Tue, 20 Feb 2007 19:35:17 +0000 (19:35 +0000)]
Update pgpass Win32 wording.

17 years agoUpdate information_schema documentation to match system tables.
Bruce Momjian [Tue, 20 Feb 2007 18:47:25 +0000 (18:47 +0000)]
Update information_schema documentation to match system tables.
Backpatch to 8.2.X.

17 years agoWording improvement.
Bruce Momjian [Tue, 20 Feb 2007 18:18:05 +0000 (18:18 +0000)]
Wording improvement.

17 years agoMore clearly document that most PostgreSQL utilities support libpq
Bruce Momjian [Tue, 20 Feb 2007 18:10:59 +0000 (18:10 +0000)]
More clearly document that most PostgreSQL utilities support libpq
environment variables.  Backpatch to 8.2.X.

17 years agoAdjust user-facing documentation to explain why we don't check pgpass
Tom Lane [Tue, 20 Feb 2007 17:47:04 +0000 (17:47 +0000)]
Adjust user-facing documentation to explain why we don't check pgpass
file permissions on Windows.

17 years agoRemove the Query structure from the executor's API. This allows us to stop
Tom Lane [Tue, 20 Feb 2007 17:32:18 +0000 (17:32 +0000)]
Remove the Query structure from the executor's API.  This allows us to stop
storing mostly-redundant Query trees in prepared statements, portals, etc.
To replace Query, a new node type called PlannedStmt is inserted by the
planner at the top of a completed plan tree; this carries just the fields of
Query that are still needed at runtime.  The statement lists kept in portals
etc. now consist of intermixed PlannedStmt and bare utility-statement nodes
--- no Query.  This incidentally allows us to remove some fields from Query
and Plan nodes that shouldn't have been there in the first place.

Still to do: simplify the execution-time range table; at the moment the
range table passed to the executor still contains Query trees for subqueries.

initdb forced due to change of stored rules.

17 years agoRemove, item not needed for Win32:
Bruce Momjian [Tue, 20 Feb 2007 15:21:42 +0000 (15:21 +0000)]
Remove, item not needed for Win32:

<  o Check .pgpass file permissions
>

17 years agoAdd comment that on Win32, we don't need to check the .pgpass file
Bruce Momjian [Tue, 20 Feb 2007 15:20:51 +0000 (15:20 +0000)]
Add comment that on Win32, we don't need to check the .pgpass file
permission, per Magnus.

17 years agoRemove "History" sections for pg_dump and pg_restore. I don't think
Bruce Momjian [Tue, 20 Feb 2007 15:01:17 +0000 (15:01 +0000)]
Remove "History" sections for pg_dump and pg_restore.  I don't think
anyone cares because the last history change was for PostgreSQL 7.1.

17 years agoSpell check on array patch.
Bruce Momjian [Tue, 20 Feb 2007 14:54:47 +0000 (14:54 +0000)]
Spell check on array patch.

17 years agoAdd for WIN32:
Bruce Momjian [Tue, 20 Feb 2007 14:48:47 +0000 (14:48 +0000)]
Add for WIN32:

<
>  o Check .pgpass file permissions

17 years agoUpdate array slice documentation to be clearer.
Bruce Momjian [Tue, 20 Feb 2007 14:34:25 +0000 (14:34 +0000)]
Update array slice documentation to be clearer.

17 years agoAdd:
Bruce Momjian [Tue, 20 Feb 2007 14:17:24 +0000 (14:17 +0000)]
Add:

> * Fix IS OF so it matches the ISO specification, and add documentation
>
>   http://archives.postgresql.org/pgsql-patches/2003-08/msg00060.php
>   http://archives.postgresql.org/pgsql-hackers/2007-02/msg00060.php

17 years agoComment-out documentation for IS OF because it doesn't conform to the
Bruce Momjian [Tue, 20 Feb 2007 14:04:50 +0000 (14:04 +0000)]
Comment-out documentation for IS OF because it doesn't conform to the
ISO SQL behavior. Backpatch removal to 8.2.X.

17 years agoIdentify schema of inherited table in psql \d when necessary.
Peter Eisentraut [Tue, 20 Feb 2007 10:23:38 +0000 (10:23 +0000)]
Identify schema of inherited table in psql \d when necessary.

by Bernd Helmle

17 years agoAdd missing OIDs to pg_proc.
Peter Eisentraut [Tue, 20 Feb 2007 10:00:25 +0000 (10:00 +0000)]
Add missing OIDs to pg_proc.

17 years agoRemove item:
Bruce Momjian [Tue, 20 Feb 2007 03:47:03 +0000 (03:47 +0000)]
Remove item:

<  o ARRAY[[1,2],[3,4]])[1] should return the same values as
<    ARRAY[[1,2],[3,4]])[1:1];
<

17 years agoDocument IS [NOT] OF, which was added in 7.3.
Bruce Momjian [Tue, 20 Feb 2007 00:25:05 +0000 (00:25 +0000)]
Document IS [NOT] OF, which was added in 7.3.

17 years agoAdd FAQ text about Makefile.custom:
Bruce Momjian [Mon, 19 Feb 2007 23:45:38 +0000 (23:45 +0000)]
Add FAQ text about Makefile.custom:

    <P><I>src/Makefile.custom</I> can be used to set environment variables,
    like <I>CUSTOM_COPT</I>, that are used for every compile.

17 years agoDocument that to_char('J'/Julian) is midnight-based, per report that
Bruce Momjian [Mon, 19 Feb 2007 22:51:44 +0000 (22:51 +0000)]
Document that to_char('J'/Julian) is midnight-based, per report that
Julian technically is noon to noon.

17 years agoWording improvement.
Bruce Momjian [Mon, 19 Feb 2007 22:10:52 +0000 (22:10 +0000)]
Wording improvement.

17 years agoUpdate PQfree() documentation to be clearer, backpatch to 8.2.X.
Bruce Momjian [Mon, 19 Feb 2007 22:06:23 +0000 (22:06 +0000)]
Update PQfree() documentation to be clearer, backpatch to 8.2.X.

17 years agoUpdate wording:
Bruce Momjian [Mon, 19 Feb 2007 21:08:48 +0000 (21:08 +0000)]
Update wording:

< * Consider allowing configuration of TOAST thresholds
> * Allow user configuration of TOAST thresholds

17 years agoAdd:
Bruce Momjian [Mon, 19 Feb 2007 21:07:25 +0000 (21:07 +0000)]
Add:

> * Consider allowing configuration of TOAST thresholds
>
>   http://archives.postgresql.org/pgsql-hackers/2007-02/msg00213.php
>

17 years agoAdd newlines.
Bruce Momjian [Mon, 19 Feb 2007 20:42:33 +0000 (20:42 +0000)]
Add newlines.

17 years agoAdd:
Bruce Momjian [Mon, 19 Feb 2007 20:41:40 +0000 (20:41 +0000)]
Add:

> * Allow UPDATEs on only non-referential integrity columns not to conflict
>   with referential integrity locks
>
>   http://archives.postgresql.org/pgsql-hackers/2007-02/msg00073.php

17 years agoAdd:
Bruce Momjian [Mon, 19 Feb 2007 20:06:56 +0000 (20:06 +0000)]
Add:

> * Allow INSERT/UPDATE ... RETURNING inside a SELECT 'FROM' clause
>
>   http://archives.postgresql.org/pgsql-general/2006-09/msg00803.php
>   http://archives.postgresql.org/pgsql-hackers/2006-10/msg00693.php
>

17 years agoClarify documentation that initdb -A or editing pg_hba.conf is required
Bruce Momjian [Mon, 19 Feb 2007 18:57:18 +0000 (18:57 +0000)]
Clarify documentation that initdb -A or editing pg_hba.conf is required
if you do not trust local users.

17 years agoClarify documentation for "day of the week" handling for to_char() and
Bruce Momjian [Mon, 19 Feb 2007 18:20:06 +0000 (18:20 +0000)]
Clarify documentation for "day of the week" handling for to_char() and
EXTRACT().

17 years agoDone:
Bruce Momjian [Mon, 19 Feb 2007 17:58:32 +0000 (17:58 +0000)]
Done:

> * -Add ISO day of week format 'ID' to to_char() where Monday = 1

17 years agoAdd "isodow" option to EXTRACT() and date_part() where Sunday = 7.
Bruce Momjian [Mon, 19 Feb 2007 17:41:39 +0000 (17:41 +0000)]
Add "isodow" option to EXTRACT() and date_part() where Sunday = 7.

17 years agoDone:
Bruce Momjian [Mon, 19 Feb 2007 17:21:24 +0000 (17:21 +0000)]
Done:

> * -Add a field 'isoyear' to extract(), based on the ISO week

17 years agoDone:
Bruce Momjian [Mon, 19 Feb 2007 16:36:17 +0000 (16:36 +0000)]
Done:

<  o Add long file support for binary pg_dump output
<
<    While Win32 supports 64-bit files, the MinGW API does not,
<    meaning we have to build an fseeko replacement on top of the
<    Win32 API, and we have to make sure MinGW handles it.  Another
<    option is to wait for the MinGW project to fix it, or use the
<    code from the LibGW32C project as a guide.
<
<    http://archives.postgresql.org/pgsql-hackers/2006-12/msg00551.php
<
>  o -Add long file support for binary pg_dump output

17 years agoFix pg_dump on win32 to properly dump files larger than 2Gb when using
Magnus Hagander [Mon, 19 Feb 2007 15:05:06 +0000 (15:05 +0000)]
Fix pg_dump on win32 to properly dump files larger than 2Gb when using
binary dump formats.

17 years agoParse pg_proc.h with entries without OIDs.
Magnus Hagander [Mon, 19 Feb 2007 14:05:42 +0000 (14:05 +0000)]
Parse pg_proc.h with entries without OIDs.

17 years agoGet rid of some old and crufty global variables in the planner. When
Tom Lane [Mon, 19 Feb 2007 07:03:34 +0000 (07:03 +0000)]
Get rid of some old and crufty global variables in the planner.  When
this code was last gone over, there wasn't really any alternative to
globals because we didn't have the PlannerInfo struct being passed all
through the planner code.  Now that we do, we can restructure things
to avoid non-reentrancy.  I'm fooling with this because otherwise I'd
have had to add another global variable for the planned compact
range table list.

17 years agoImprove plpgsql's error message when a datatype declaration is omitted.
Tom Lane [Mon, 19 Feb 2007 03:18:51 +0000 (03:18 +0000)]
Improve plpgsql's error message when a datatype declaration is omitted.
Per example from Jeff Ross.

17 years agoPut function expressions and values lists into FunctionScan and ValuesScan
Tom Lane [Mon, 19 Feb 2007 02:23:12 +0000 (02:23 +0000)]
Put function expressions and values lists into FunctionScan and ValuesScan
plan nodes, so that the executor does not need to get these items from
the range table at runtime.  This will avoid needing to include these
fields in the compact range table I'm expecting to make the executor use.

17 years agoFix portal management code to support non-default command completion tags for
Tom Lane [Sun, 18 Feb 2007 19:49:25 +0000 (19:49 +0000)]
Fix portal management code to support non-default command completion tags for
portals using PORTAL_UTIL_SELECT strategy.  This is currently significant only
for FETCH queries, which are supposed to include a count in the tag.  Seems
it's been broken since 7.4, but nobody noticed before Knut Lehre.

17 years agoDocument limitation of SPI_execute() using multiple queries in a single
Bruce Momjian [Sun, 18 Feb 2007 01:47:40 +0000 (01:47 +0000)]
Document limitation of SPI_execute() using multiple queries in a single
string.

17 years agoUpdate wording:
Bruce Momjian [Sun, 18 Feb 2007 01:34:35 +0000 (01:34 +0000)]
Update wording:

<   Currently, ALTER USER and ALTER DATABASE support per-user and
>   Currently ALTER USER and ALTER DATABASE support per-user and
<    Currently, subtracting one date from another that crosses a
>    Currently subtracting one date from another that crosses a
<   Currently, SQL-language functions can only refer to parameters via $1, etc
>   Currently SQL-language functions can only refer to dollar parameters,
>   e.g. $1
<   Currently, queries prepared via the libpq API are planned on first
>   Currently queries prepared via the libpq API are planned on first
<    Currently, SET <tab> causes a database lookup to check all
>    Currently SET <tab> causes a database lookup to check all
<    Currently, all statement results are transferred to the libpq
>    Currently all statement results are transferred to the libpq

17 years agoAdd:
Bruce Momjian [Sun, 18 Feb 2007 01:32:55 +0000 (01:32 +0000)]
Add:

* Allow SQL-language functions to reference parameters by parameter name

  Currently SQL-language functions can only refer to parameters via $1, etc

17 years agoUpdate "Conflicting lock modes" to show as conflict, add
Bruce Momjian [Sun, 18 Feb 2007 01:21:49 +0000 (01:21 +0000)]
Update "Conflicting lock modes" to show as conflict, add
current/requested headings, add link to table from text.

17 years agoAdd code so that when COPY_PARSE_PLAN_TREES is defined, the copy and
Tom Lane [Sat, 17 Feb 2007 19:33:32 +0000 (19:33 +0000)]
Add code so that when COPY_PARSE_PLAN_TREES is defined, the copy and
equal functions are checked for raw parse trees as well as post-analysis
trees.  This was never very important before, but the upcoming plan cache
control module will need to be able to do copyObject() on raw parse trees.

17 years agoRemove rint() for to_char MS and US output. We can't us rint() because
Bruce Momjian [Sat, 17 Feb 2007 03:11:32 +0000 (03:11 +0000)]
Remove rint() for to_char MS and US output.  We can't us rint() because
we can't overflow to the next higher units, and we might print the lower
units for MS.

17 years agoAdd:
Bruce Momjian [Sat, 17 Feb 2007 02:01:02 +0000 (02:01 +0000)]
Add:

> * Allow holdable cursors in SPI

17 years agoCleanup of to_char() patch.
Bruce Momjian [Sat, 17 Feb 2007 01:51:42 +0000 (01:51 +0000)]
Cleanup of to_char() patch.

Brendan Jurd

17 years agoAdd:
Bruce Momjian [Sat, 17 Feb 2007 01:35:41 +0000 (01:35 +0000)]
Add:
>
>  o Allow row and record variables to be set to NULL constants,
>    and allow NULL tests on such variables
>
>    Because a row is not scalar, do not allow assignment
>    from NULL-valued scalars.

17 years agoRemove installation mention that integer timestamps is less stable that
Bruce Momjian [Sat, 17 Feb 2007 01:26:24 +0000 (01:26 +0000)]
Remove installation mention that integer timestamps is less stable that
floating point.

17 years agoCreate AVG() aggregates for int8 and NUMERIC which do not compute X^2,
Bruce Momjian [Sat, 17 Feb 2007 00:55:58 +0000 (00:55 +0000)]
Create AVG() aggregates for int8 and NUMERIC which do not compute X^2,
as a performance enhancement.

Mark Kirkwood

17 years agoTeach find_nonnullable_rels to handle OR cases: if every arm of an OR
Tom Lane [Fri, 16 Feb 2007 23:32:08 +0000 (23:32 +0000)]
Teach find_nonnullable_rels to handle OR cases: if every arm of an OR
forces a particular relation nonnullable, then we can say that the OR does.
This is worth a little extra trouble since it may allow reduction of
outer joins to plain joins.

17 years agoAdd:
Bruce Momjian [Fri, 16 Feb 2007 22:20:02 +0000 (22:20 +0000)]
Add:

>  o Consider reducing on-disk varlena length from four to two
>    because a heap row cannot be more than 64k in length

17 years agoFix new RI operator selection code to do the right thing when working with
Tom Lane [Fri, 16 Feb 2007 22:04:02 +0000 (22:04 +0000)]
Fix new RI operator selection code to do the right thing when working with
an opclass for a generic type such as ANYARRAY.  The original coding failed
to check that PK and FK columns were of the same array type.  Per discussion
with Tom Dunstan.  Also, make the code a shade more readable by not trying
to economize on variables.

17 years agoReduce the amount of memory "clobbered" for every process title change,
Bruce Momjian [Fri, 16 Feb 2007 21:34:04 +0000 (21:34 +0000)]
Reduce the amount of memory "clobbered" for every process title change,
on platforms that need this.  This is done by only writing past the
previously stored message, if it was longer.

17 years agoAdjust the definition of is_pushed_down so that it's always true for INNER
Tom Lane [Fri, 16 Feb 2007 20:57:19 +0000 (20:57 +0000)]
Adjust the definition of is_pushed_down so that it's always true for INNER
JOIN quals, just like WHERE quals, even if they reference every one of the
join's relations.  Now that we can reorder outer and inner joins, it's
possible for such a qual to end up being assigned to an outer join plan node,
and we mustn't have it treated as a join qual rather than a filter qual for
the node.  (If it were, the join could produce null-extended rows that it
shouldn't.)  Per bug report from Pelle Johansson.

17 years agoBetter fix for determining minimum and maximum int64 values that doesn't
Peter Eisentraut [Fri, 16 Feb 2007 18:37:43 +0000 (18:37 +0000)]
Better fix for determining minimum and maximum int64 values that doesn't
require stdint.h and works for "busted" int64.

17 years agoInstall a more correct fix in the timestamp and timestamptz regression tests:
Alvaro Herrera [Fri, 16 Feb 2007 17:49:15 +0000 (17:49 +0000)]
Install a more correct fix in the timestamp and timestamptz regression tests:
remove duplicated tests in timestamp, and complete timestamptz with the tests
that were missing to more closely mirror timestamp.

17 years agoCode review for SSLKEY patch.
Tom Lane [Fri, 16 Feb 2007 17:07:00 +0000 (17:07 +0000)]
Code review for SSLKEY patch.

17 years agoFix markup, spelling, grammar, and explanations for SSLKEY patch.
Tom Lane [Fri, 16 Feb 2007 16:37:29 +0000 (16:37 +0000)]
Fix markup, spelling, grammar, and explanations for SSLKEY patch.

17 years agoFix the timestamptz test problem, by moving the tests that use the
Alvaro Herrera [Fri, 16 Feb 2007 15:42:42 +0000 (15:42 +0000)]
Fix the timestamptz test problem, by moving the tests that use the
timestamp_tbl table into the timestamp test.  Also, restore a test that
used to exist as a valid test in the timestamptz test.

17 years agoFix // comment
Peter Eisentraut [Fri, 16 Feb 2007 10:55:47 +0000 (10:55 +0000)]
Fix // comment

17 years agoAdd stdint.h include
Peter Eisentraut [Fri, 16 Feb 2007 10:42:31 +0000 (10:42 +0000)]
Add stdint.h include

17 years agoFunctions for mapping table data and table schemas to XML (a.k.a. XML export)
Peter Eisentraut [Fri, 16 Feb 2007 07:46:55 +0000 (07:46 +0000)]
Functions for mapping table data and table schemas to XML (a.k.a. XML export)

17 years agoRemove extra character erroneously added.
Bruce Momjian [Fri, 16 Feb 2007 03:56:45 +0000 (03:56 +0000)]
Remove extra character erroneously added.

17 years agoRemove tabs from SGML files to help tag alingment and improve
Bruce Momjian [Fri, 16 Feb 2007 03:50:29 +0000 (03:50 +0000)]
Remove tabs from SGML files to help tag alingment and improve
detection of tabs are added in the future.

17 years agoFix another problem in 8.2 changes that allowed "one-time" qual conditions to
Tom Lane [Fri, 16 Feb 2007 03:49:04 +0000 (03:49 +0000)]
Fix another problem in 8.2 changes that allowed "one-time" qual conditions to
be checked at plan levels below the top; namely, we have to allow for Result
nodes inserted just above a nestloop inner indexscan.  Should think about
using the general Param mechanism to pass down outer-relation variables, but
for the moment we need a back-patchable solution.  Per report from Phil Frost.

17 years agoAdd two new format fields for use with to_char(), to_date() and
Bruce Momjian [Fri, 16 Feb 2007 03:39:46 +0000 (03:39 +0000)]
Add two new format fields for use with to_char(), to_date() and
to_timestamp():
    - ID for day-of-week
    - IDDD for day-of-year

This makes it possible to convert ISO week dates to and from text
fully represented in either week ('IYYY-IW-ID') or day-of-year
('IYYY-IDDD') format.

I have also added an 'isoyear' field for use with extract / date_part.

Brendan Jurd

17 years agoSSL improvements:
Bruce Momjian [Fri, 16 Feb 2007 02:59:41 +0000 (02:59 +0000)]
SSL improvements:

o read global SSL configuration file
o add GUC "ssl_ciphers" to control allowed ciphers
o add libpq environment variable PGSSLKEY to control SSL hardware keys

Victor B. Wagner

17 years agoRemove useless database name from bootstrap argument processing (including
Alvaro Herrera [Fri, 16 Feb 2007 02:10:07 +0000 (02:10 +0000)]
Remove useless database name from bootstrap argument processing (including
startup and bgwriter processes), and the -y flag.  It's not used anywhere.

17 years agoRestructure code that is responsible for ensuring that clauseless joins are
Tom Lane [Fri, 16 Feb 2007 00:14:01 +0000 (00:14 +0000)]
Restructure code that is responsible for ensuring that clauseless joins are
considered when it is necessary to do so because of a join-order restriction
(that is, an outer-join or IN-subselect construct).  The former coding was a
bit ad-hoc and inconsistent, and it missed some cases, as exposed by Mario
Weilguni's recent bug report.  His specific problem was that an IN could be
turned into a "clauseless" join due to constant-propagation removing the IN's
joinclause, and if the IN's subselect involved more than one relation and
there was more than one such IN linking to the same upper relation, then the
only valid join orders involve "bushy" plans but we would fail to consider the
specific paths needed to get there.  (See the example case added to the join
regression test.)  On examining the code I wonder if there weren't some other
problem cases too; in particular it seems that GEQO was defending against a
different set of corner cases than the main planner was.  There was also an
efficiency problem, in that when we did realize we needed a clauseless join
because of an IN, we'd consider clauseless joins against every other relation
whether this was sensible or not.  It seems a better design is to use the
outer-join and in-clause lists as a backup heuristic, just as the rule of
joining only where there are joinclauses is a heuristic: we'll join two
relations if they have a usable joinclause *or* this might be necessary to
satisfy an outer-join or IN-clause join order restriction.  I refactored the
code to have just one place considering this instead of three, and made sure
that it covered all the cases that any of them had been considering.

Backpatch as far as 8.1 (which has only the IN-clause form of the disease).
By rights 8.0 and 7.4 should have the bug too, but they accidentally fail
to fail, because the joininfo structure used in those releases preserves some
memory of there having once been a joinclause between the inner and outer
sides of an IN, and so it leads the code in the right direction anyway.
I'll be conservative and not touch them.

17 years agoRestructure autovacuum in two processes: a dummy process, which runs
Alvaro Herrera [Thu, 15 Feb 2007 23:23:23 +0000 (23:23 +0000)]
Restructure autovacuum in two processes: a dummy process, which runs
continuously, and requests vacuum runs of "autovacuum workers" to postmaster.
The workers do the actual vacuum work.  This allows for future improvements,
like allowing multiple autovacuum jobs running in parallel.

For now, the code keeps the original behavior of having a single autovac
process at any time by sleeping until the previous worker has finished.

17 years agoAdd ORDER BY to a query on information_schema.views, to avoid possible
Tom Lane [Thu, 15 Feb 2007 05:05:03 +0000 (05:05 +0000)]
Add ORDER BY to a query on information_schema.views, to avoid possible
platform-specific result ordering.  Per buildfarm results.

17 years agoRepair oversight in 8.2 change that improved the handling of "pseudoconstant"
Tom Lane [Thu, 15 Feb 2007 03:07:13 +0000 (03:07 +0000)]
Repair oversight in 8.2 change that improved the handling of "pseudoconstant"
WHERE clauses.  createplan.c is now willing to stick a gating Result node
almost anywhere in the plan tree, and in particular one can wind up directly
underneath a MergeJoin node.  This means it had better be willing to handle
Mark/Restore.  Fortunately, that's trivial in such cases, since we can just
pass off the call to the input node (which the planner has previously ensured
can handle Mark/Restore).  Per report from Phil Frost.

17 years agoMake it possible to build with integer datetimes in msvc, and enable by default.
Magnus Hagander [Wed, 14 Feb 2007 21:02:07 +0000 (21:02 +0000)]
Make it possible to build with integer datetimes in msvc, and enable by default.

17 years agoAdd:
Bruce Momjian [Wed, 14 Feb 2007 21:00:17 +0000 (21:00 +0000)]
Add:

>
>  o Use LC_TIME for localized weekday/month names, rather than
>    LC_MESSAGES
>
>    http://archives.postgresql.org/pgsql-hackers/2006-11/msg00390.php