Peter Eisentraut [Thu, 19 Feb 2009 10:07:58 +0000 (10:07 +0000)]
Remove croak and Perl_croak from gettext triggers. While we could
selectively mark up their arguments for translation, the Perl xsubpp tool
generates a bunch of additional Perl_croak calls that we cannot control,
so we'd be creating a confusing mix of translated and untranslated messages
of a similar kind. This is something that might deserve a more
comprehensive solution later.
Also remove _ from gettext triggers, because it wasn't used.
Use SPI.c instead of SPI.xs as source file for xgettext, because the .xs
format isn't really supported in xgettext.
Increase NUM_AUXILIARY_PROCS, now that the startup process can co-exist
with other auxiliary processes for a short period. As witnessed by
buildfarm member dungbeetle.
Tom Lane [Wed, 18 Feb 2009 19:23:26 +0000 (19:23 +0000)]
Remove the special cases to prevent minus-zero results in float4 and float8
unary minus operators. We weren't attempting to prevent minus zero anywhere
else; in view of our gradual trend to make the float datatypes more IEEE
standard compliant, we should allow minus zero here rather than disallow it
elsewhere.
We don't, however, expect that all platforms will produce minus zero, so
we need to adjust the one affected regression test to allow both results.
Per discussion of bug #4660.
(In passing, clean up a couple other minor infelicities in float.c.)
Start background writer during archive recovery. Background writer now performs
its usual buffer cleaning duties during archive recovery, and it's responsible
for performing restartpoints.
This requires some changes in postmaster. When the startup process has done
all the initialization and is ready to start WAL redo, it signals the
postmaster to launch the background writer. The postmaster is signaled again
when the point in recovery is reached where we know that the database is in
consistent state. Postmaster isn't interested in that at the moment, but
that's the point where we could let other backends in to perform read-only
queries. The postmaster is signaled third time when the recovery has ended,
so that postmaster knows that it's safe to start accepting connections.
The startup process now traps SIGTERM, and performs a "clean" shutdown. If
you do a fast shutdown during recovery, a shutdown restartpoint is performed,
like a shutdown checkpoint, and postmaster kills the processes cleanly. You
still have to continue the recovery at next startup, though.
Currently, the background writer is only launched during archive recovery.
We could launch it during crash recovery as well, but it seems better to keep
that codepath as simple as possible, for the sake of robustness. And it
couldn't do any restartpoints during crash recovery anyway, so it wouldn't be
that useful.
log_restartpoints is gone. Use log_checkpoints instead. This is yet to be
documented.
This whole operation is a pre-requisite for Hot Standby, but has some value of
its own whether the hot standby patch makes 8.4 or not.
Tom Lane [Sun, 15 Feb 2009 20:16:21 +0000 (20:16 +0000)]
Teach the planner to treat a partial unique index as proving a variable is
unique for a particular query, if the index predicate is satisfied. This
requires a bit of reordering of operations so that we check the predicates
before doing any selectivity estimates, but shouldn't really cause any
noticeable slowdown. Per a comment from Michal Politowski.
Tom Lane [Sat, 14 Feb 2009 20:48:36 +0000 (20:48 +0000)]
A couple of marginal performance hacks for the information_schema views:
replace the old recursive-SQL-function implementation of _pg_keysequal()
with use of the built-in array containment operators, and change
table_constraints' UNION to UNION ALL. Per discussion with Octavio Alvarez.
initdb not forced since this doesn't affect results, but you'd need to
initdb or reload the information_schema to see the new definitions.
Tom Lane [Fri, 13 Feb 2009 17:12:04 +0000 (17:12 +0000)]
Fix UNLISTEN to fall out quickly if the current backend has never executed
any LISTEN command. This is more important than it used to be because
DISCARD ALL invokes UNLISTEN. Connection-pooled applications making heavy
use of DISCARD ALL were seeing significant contention for pg_listener,
as reported by Matteo Beccati. It seems unlikely that clients using LISTEN
would use pooled connections, so this simple tweak seems sufficient,
especially since the pg_listener implementation is slated to go away soon
anyway.
Back-patch to 8.3, where DISCARD ALL was introduced.
Peter Eisentraut [Thu, 12 Feb 2009 15:11:44 +0000 (15:11 +0000)]
The Czech (cs_CZ) and Slovak (sk_SK) locales sort numbers after letters,
instead of vice versa. Update the regression test expectations to support
that. In the plpgsql test, adjust the test data so that this isn't an
issue. In the char and varchar tests, add new expected files.
Magnus Hagander [Thu, 12 Feb 2009 12:53:34 +0000 (12:53 +0000)]
Don't call SetEnvironmentVariable() when removing an environment variable,
as this seems to crash on at least some versions of MingW. Our current usage
of this function does not require it, so it should be ok to ignore.
Tom Lane [Wed, 11 Feb 2009 21:11:16 +0000 (21:11 +0000)]
Change ALTER TABLE SET WITHOUT OIDS to rewrite the whole table to physically
get rid of the OID column. This eliminates the problem discovered by Heikki
back in November that 8.4's suppression of "unnecessary" junk filtering in
INSERT/SELECT could lead to an Assert failure, or storing of oids into a table
that shouldn't have them if Asserts are off. While that particular problem
could have been solved in other ways, it seems likely to be just a forerunner
of things to come if we continue to allow tables to contain rows that disagree
with the pg_class.relhasoids setting. It's better to make this operation slow
than to sacrifice performance or risk bugs in more common code paths.
Also, add ALTER TABLE SET WITH OIDS to rewrite the table to add oids.
This was a bit more controversial, but in view of the very small amount of
extra code needed given the current ALTER TABLE infrastructure, it seems best
to eliminate the asymmetry in features.
Tom Lane [Wed, 11 Feb 2009 20:02:40 +0000 (20:02 +0000)]
Tweak configure to attempt to add -qnoansialias to CFLAGS whenever running
on AIX with a non-gcc compiler. The previous coding would do this only if
CC was exactly "xlc"; which is a bad idea, as demonstrated by trouble report
from Mihai Criveti.
Peter Eisentraut [Wed, 11 Feb 2009 14:03:42 +0000 (14:03 +0000)]
Only unset the locale environment when --no-locale is used (the way it was
presumably designed, but didn't act). This allows running the temp install
tests in a non-C locale, thus exercising users' real environments better.
Document how to change locales for test runs.
Tom Lane [Sun, 8 Feb 2009 18:02:27 +0000 (18:02 +0000)]
Ensure that INSERT ... SELECT into a table with OIDs never copies row OIDs
from the source table. This could never happen anyway before 8.4 because
the executor invariably applied a "junk filter" to rows due to be inserted;
but now that we skip doing that when it's not necessary, the case can occur.
Problem noted 2008-11-27 by KaiGai Kohei, though I misunderstood what he
was on about at the time (the opacity of the patch he proposed didn't help).
Bruce Momjian [Sat, 7 Feb 2009 14:16:46 +0000 (14:16 +0000)]
Fix to_timestamp() to not require upper/lower case matching for meridian
designations (AM/PM). Also separate out matching of a meridian with
periods (e.g. A.M.) and with those without.
Tom Lane [Fri, 6 Feb 2009 23:43:24 +0000 (23:43 +0000)]
Fix cost_mergejoin's failure to adjust for rescanning of non-unique merge join
keys when considering a semi or anti join. This requires estimating the
selectivity of the merge qual as though it were a regular inner join condition.
To allow caching both that and the real outer-join-aware selectivity, split
RestrictInfo.this_selec into two fields.
This fixes one of the problems reported by Kevin Grittner.
Tom Lane [Fri, 6 Feb 2009 21:15:12 +0000 (21:15 +0000)]
Clean up some loose ends from the column privileges patch: add
has_column_privilege and has_any_column_privilege SQL functions; fix the
information_schema views that are supposed to pay attention to column
privileges; adjust pg_stats to show stats for any column you have select
privilege on; and fix COPY to allow copying a subset of columns if the user
has suitable per-column privileges for all the columns.
To improve efficiency of some of the information_schema views, extend the
has_xxx_privilege functions to allow inquiring about the OR of a set of
privileges in just one call. This is just exposing capability that already
existed in the underlying aclcheck routines.
In passing, make the information_schema views report the owner's own
privileges as being grantable, since Postgres assumes this even when the grant
option bit is not set in the ACL. This is a longstanding oversight.
Also, make the new has_xxx_privilege functions for foreign data objects follow
the same coding conventions used by the older ones.
Tom Lane [Thu, 5 Feb 2009 01:24:55 +0000 (01:24 +0000)]
Fix an old corner-case error in match_unsorted_outer(): don't consider
the cheapest-total inner path as a new candidate while truncating the
sort key list, if it already matched the full sort key list. This is
too much of a corner case to be worth back-patching, since it's unusual
for the cheapest total path to be sorted, and anyway no real harm is
done (except in JOIN_SEMI/ANTI cases where cost_mergejoin is a bit
broken at the moment). But it wasn't behaving as intended, so fix it.
Noted while examining a test case from Kevin Grittner. This error doesn't
explain his issue, but it does explain why "set enable_seqscan = off"
seemed to reproduce it for me.
Bruce Momjian [Wed, 4 Feb 2009 16:05:50 +0000 (16:05 +0000)]
Update read committed documentation to better explain undesirable
behavior of concurrent commands in cases where rows are being added and
removed from matching query criteria.
Tom Lane [Tue, 3 Feb 2009 00:59:26 +0000 (00:59 +0000)]
Add a failure check for syslogger's use of _beginthreadex(), and remove
unnecessary thread address output parameter, to make this code look more
like that in pg_restore.
Tom Lane [Mon, 2 Feb 2009 20:42:57 +0000 (20:42 +0000)]
Document the longstanding behavior of LIMIT NULL and OFFSET NULL,
per gripe from David Wheeler that this was mentioned nowhere.
In passing, editorialize a bit on the description of the SQL:2008
equivalent syntax.
Tom Lane [Mon, 2 Feb 2009 20:25:38 +0000 (20:25 +0000)]
Fix plpgsql to not treat INSERT INTO as an INTO-variables clause anywhere
in the string, not just at the start. Per bug #4629 from Martin Blazek.
Back-patch to 8.2; prior versions don't have the problem, at least not in
the reported case, because they don't try to recognize INTO in non-SELECT
statements. (IOW, this is really fallout from the RETURNING patch.)
Andrew Dunstan [Mon, 2 Feb 2009 20:07:37 +0000 (20:07 +0000)]
Provide for parallel restoration from a custom format archive. Each data and
post-data step is run in a separate worker child (a thread on Windows, a child
process elsewhere) up to the concurrent number specified by the new pg_restore
command-line --multi-thread | -m switch.
Alvaro Herrera [Mon, 2 Feb 2009 19:31:40 +0000 (19:31 +0000)]
Allow reloption names to have qualifiers, initially supporting a TOAST
qualifier, and add support for this in pg_dump.
This allows TOAST tables to have user-defined fillfactor, and will also
enable us to move the autovacuum parameters to reloptions without taking
away the possibility of setting values for TOAST tables.
Check that connection limit is within valid range. IOW, not < -1.
It's missing in older versions too, but it doesn't seem worth
back-porting. All negative are just harmlessly treated as "no limit", and
tightening the check might even brake an application that relies on it.
Tom Lane [Thu, 29 Jan 2009 19:59:59 +0000 (19:59 +0000)]
Update time zone data files to tzdata release 2009a: introduces Asia/Kathmandu
as the preferred spelling of that zone name, corrects historical DST
information for Switzerland and Cuba.
Tom Lane [Thu, 29 Jan 2009 19:23:42 +0000 (19:23 +0000)]
Replace argument-checking Asserts with regular test-and-elog checks in all
encoding conversion functions. These are not can't-happen cases because
it's possible to create a conversion with the wrong conversion function
for the specified encoding pair. That would lead to an Assert crash in
an Assert-enabled build, or incorrect conversion otherwise, neither of
which is desirable. This would be a DOS issue if production databases
were customarily built with asserts enabled, but fortunately that's not so.
Per an observation by Heikki.
Teodor Sigaev [Wed, 28 Jan 2009 18:31:32 +0000 (18:31 +0000)]
Fix bug with multiple evaluation of tsearch2 compatibility trigger, trigger
data should be restored.
Backpatch only for 8.3 because previous versions haven't such layer.
Magnus Hagander [Wed, 28 Jan 2009 15:06:47 +0000 (15:06 +0000)]
Go over all OpenSSL return values and make sure we compare them
to the documented API value. The previous code got it right as
it's implemented, but accepted too much/too little compared to
the API documentation.
Alvaro Herrera [Mon, 26 Jan 2009 19:41:06 +0000 (19:41 +0000)]
Allow extracting and parsing of reloptions from a bare pg_class tuple, and
refactor the relcache code that used to do that. This allows other callers
(particularly autovacuum) to do the same without necessarily having to open
and lock a table.
Tom Lane [Fri, 23 Jan 2009 19:58:06 +0000 (19:58 +0000)]
Tweak the existing special case for AIX in pg_getaddrinfo_all() to handle
yet another failure case in AIX's getaddrinfo(). Per report and patch
by Andrew Chernow.
Put back fast-path for the case that there's no backup blocks in
RestoreBkpBlocks. Went missing in my recent refactoring patch, as pointed
out by Simon's hot standby patch.
Peter Eisentraut [Thu, 22 Jan 2009 22:27:13 +0000 (22:27 +0000)]
Recreate ecpg_config.h with a stamp file created by config.status, same
way pg_config.h is handled. This avoids reruns of config.status on every
build, if configure has been rerun.
Only skip pages marked as clean in the visibility map, if the last 32
pages were marked as clean as well. The idea is to avoid defeating OS
readahead by skipping a page here and there, and also makes it less likely
that we miss an opportunity to advance relfrozenxid, for the sake of only
a few skipped pages.