Bruce Momjian [Wed, 4 Apr 2007 17:28:11 +0000 (17:28 +0000)]
Remove plpython casts C source code casts:
It removes last remaining casts inside struct definitions.
Such usage is bad practice, as it hides problems from compiler.
Reason for the cast is popular practice in some circles
to define functions as foo(MyObj *) instead of foo(PyObject *)
thus avoiding a local variable inside functions and make
direct calling easier. As pl/python does not use such style,
the casts were unnecessary from the start.
Tom Lane [Tue, 3 Apr 2007 16:34:36 +0000 (16:34 +0000)]
Remove the CheckpointStartLock in favor of having backends show whether they
are in their commit critical sections via flags in the ProcArray. Checkpoint
can watch the ProcArray to determine when it's safe to proceed. This is
a considerably better solution to the original problem of race conditions
between checkpoint and transaction commit: it speeds up commit, since there's
one less lock to fool with, and it prevents the problem of checkpoint being
delayed indefinitely when there's a constant flow of commits. Heikki, with
some kibitzing from Tom.
Tom Lane [Tue, 3 Apr 2007 04:14:26 +0000 (04:14 +0000)]
Decouple the values of TOAST_TUPLE_THRESHOLD and TOAST_MAX_CHUNK_SIZE.
Add the latter to the values checked in pg_control, since it can't be changed
without invalidating toast table content. This commit in itself shouldn't
change any behavior, but it lays some necessary groundwork for experimentation
with these toast-control numbers.
Note: while TOAST_TUPLE_THRESHOLD can now be changed without initdb, some
thought still needs to be given to needs_toast_table() in toasting.c before
unleashing random changes.
Bruce Momjian [Tue, 3 Apr 2007 02:17:11 +0000 (02:17 +0000)]
Update:
< * Add idle_timeout GUC so locks are not held for log periods of time
> * Add transaction_idle_timeout GUC so locks are not held for long
> periods of time
Bruce Momjian [Mon, 2 Apr 2007 22:49:03 +0000 (22:49 +0000)]
Add URL for:
* Improve speed with indexes
For large table adjustments during VACUUM FULL, it is faster to cluster
or reindex rather than update the index. Also, index updates can bloat
the index.
Bruce Momjian [Mon, 2 Apr 2007 22:46:10 +0000 (22:46 +0000)]
Add:
> o Have timestamp subtraction not call justify_hours()?
>
> http://archives.postgresql.org/pgsql-sql/2006-10/msg00059.php
>
< o Add overflow checking to timestamp and interval arithmetic
> o Add overflow checking to timestamp and interval arithmetic
Bruce Momjian [Mon, 2 Apr 2007 22:20:53 +0000 (22:20 +0000)]
Allow NOTIFY/LISTEN/UNLISTEN to only take relation names, not
schema.relation, because the notify code only honors the relation name.
schema.relation will now generate a syntax error.
Bruce Momjian [Mon, 2 Apr 2007 21:28:10 +0000 (21:28 +0000)]
Pl/python -> Pl/PythonU:
< o Add table function support to pltcl, plpython
> o Add table function support to pltcl, plpythonu
< o Add PL/Python tracebacks
> o Add PL/PythonU tracebacks
< o Allow PL/Python to return boolean rather than 1/0
> o Allow PL/PythonU to return boolean rather than 1/0
Bruce Momjian [Mon, 2 Apr 2007 20:10:16 +0000 (20:10 +0000)]
Add URL for:
o Allow RETURN to return row or record functions
> o Allow RETURN to return row or record functions
> http://archives.postgresql.org/pgsql-patches/2006-08/msg00397.php
> http://archives.postgresql.org/pgsql-hackers/2006-09/msg00388.php
Tom Lane [Mon, 2 Apr 2007 18:49:29 +0000 (18:49 +0000)]
Fix check_sql_fn_retval to allow the case where a SQL function declared to
return void ends with a SELECT, if that SELECT has a single result that is
also of type void. Without this, it's hard to write a void function that
calls another void function. Per gripe from Peter.
Bruce Momjian [Mon, 2 Apr 2007 17:18:44 +0000 (17:18 +0000)]
Done:
< * Support a data type with specific enumerated values (ENUM)
<
< http://archives.postgresql.org/pgsql-hackers/2006-08/msg00979.php
<
> * -Support a data type with specific enumerated values (ENUM)
Tom Lane [Mon, 2 Apr 2007 03:49:42 +0000 (03:49 +0000)]
Support enum data types. Along the way, use macros for the values of
pg_type.typtype whereever practical. Tom Dunstan, with some kibitzing
from Tom Lane.
Tom Lane [Fri, 30 Mar 2007 00:12:59 +0000 (00:12 +0000)]
Fix oversight in coding of _bt_start_vacuum: we can't assume that the LWLock
will be released by transaction abort before _bt_end_vacuum gets called.
If either of these "can't happen" errors actually happened, we'd freeze up
trying to acquire an already-held lock. Latest word is that this does
not explain Martin Pitt's trouble report, but it still looks like a bug.
Bruce Momjian [Thu, 29 Mar 2007 20:23:56 +0000 (20:23 +0000)]
Update item:
o Add more logical syntax CLUSTER table USING index;
< o Add more logical syntax CLUSTER table ORDER BY index;
> o Add more logical syntax CLUSTER table USING index;
Tom Lane [Thu, 29 Mar 2007 19:10:10 +0000 (19:10 +0000)]
exec_parse_message neglected to copy parameter type array into the
required memory context when handling client-specified parameter types
for an unnamed statement. Per report from Kris Jurka.
Magnus Hagander [Thu, 29 Mar 2007 15:30:52 +0000 (15:30 +0000)]
Make ECPG regression tests use native threading instead of pthreads, now that
ecpglib supports it.
Change configure (patch from Bruce) and msvc build system to no longer require
pthreads on win32, since all parts of postgresql can be thread-safe using the
native platform functions.
Tom Lane [Thu, 29 Mar 2007 00:15:39 +0000 (00:15 +0000)]
Teach CLUSTER to skip writing WAL if not needed (ie, not using archiving)
--- Simon.
Also, code review and cleanup for the previous COPY-no-WAL patches --- Tom.
Neil Conway [Wed, 28 Mar 2007 22:59:37 +0000 (22:59 +0000)]
Code cleanup for the new regexp UDFs: we can hardcode the OID and some
properties of the "text" type, and then simplify the code accordingly.
Patch from Jeremy Drake.
Alvaro Herrera [Wed, 28 Mar 2007 22:17:12 +0000 (22:17 +0000)]
Add the "recheck" logic to autovacuum worker code. The worker first builds
its table list and then rechecks pgstat before vacuuming each table to
verify that no one has vacuumed the table in the meantime.
In the current autovacuum world this only means that a worker will not
vacuum a table that a user has vacuumed manually after the worker started.
When support for multiple autovacuum workers is introduced, this will reduce
the probability of simultaneous workers on the same database doing redundant
work.
Tom Lane [Tue, 27 Mar 2007 23:21:12 +0000 (23:21 +0000)]
Fix array coercion expressions to ensure that the correct volatility is
seen by code inspecting the expression. The best way to do this seems
to be to drop the original representation as a function invocation, and
instead make a special expression node type that represents applying
the element-type coercion function to each array element. In this way
the element function is exposed and will be checked for volatility.
Per report from Guillaume Smet.
Alvaro Herrera [Tue, 27 Mar 2007 20:36:03 +0000 (20:36 +0000)]
Cosmetic changes: rename some struct fields, and move the fetching of pgstat
table entries to a separate routine. Don't pass the pgstat database entry to
do_autovacuum; rather, have it fetch it by itself.
Bruce Momjian [Tue, 27 Mar 2007 17:27:01 +0000 (17:27 +0000)]
Remove item due to lack of interest:
< * %Add pg_get_acldef(), pg_get_typedefault(), pg_get_attrdef(),
< pg_get_tabledef(), pg_get_domaindef(), pg_get_functiondef()
<
< These would be for application use, not for use by pg_dump.
<
Bruce Momjian [Tue, 27 Mar 2007 15:36:40 +0000 (15:36 +0000)]
Add:
>
> * Allow configuration of backend priorities via the operating system
>
> Though backend priorities make priority inversion during lock
> waits possible, research shows that this is not a huge problem.
> http://archives.postgresql.org/pgsql-general/2007-02/msg00493.php
Bruce Momjian [Mon, 26 Mar 2007 23:47:53 +0000 (23:47 +0000)]
Add:
> * Increase the number of rows recognized as dead during multi-statement
> transactions
>
> http://archives.postgresql.org/pgsql-patches/2007-03/msg00358.php
>
Tom Lane [Mon, 26 Mar 2007 16:58:41 +0000 (16:58 +0000)]
Allow non-superuser database owners to create procedural languages.
A DBA is allowed to create a language in his database if it's marked
"tmpldbacreate" in pg_pltemplate. The factory default is that this is set
for all standard trusted languages, but of course a superuser may adjust
the settings. In service of this, add the long-foreseen owner column to
pg_language; renaming, dropping, and altering owner of a PL now follow
normal ownership rules instead of being superuser-only.
Jeremy Drake, with some editorialization by Tom Lane.
Tom Lane [Mon, 26 Mar 2007 02:37:17 +0000 (02:37 +0000)]
Add -lcrypto as one of the possible link dependencies of libkrb5.
Per report from Jim Rosenberg. This possibly should get back-patched,
but I'm a bit suspicious of it still because of the lack of prior reports.
Tom Lane [Mon, 26 Mar 2007 01:41:57 +0000 (01:41 +0000)]
Remove assertion that constraint_exclusion risks wrong answers if
table constraints are changed; this is no longer true now that we
have a plan invalidation mechanism.
Tom Lane [Mon, 26 Mar 2007 00:36:19 +0000 (00:36 +0000)]
Fix plancache's invalidation callback to do the right thing for a SI
reset event, namely invalidate everything. This oversight probably
explains the rare failures that some buildfarm machines have been
showing for the plancache regression test.
Tom Lane [Sun, 25 Mar 2007 23:42:43 +0000 (23:42 +0000)]
Make _SPI_execute_plan pass the query source string down to ProcessUtility
if possible. I had left this undone in the first pass at the API change
for ProcessUtility, but forgot to revisit it after the plancache changes
made it possible to do it.
Tom Lane [Sun, 25 Mar 2007 23:27:59 +0000 (23:27 +0000)]
Remove the prohibition on executing cursor commands through SPI_execute.
Vadim had included this restriction in the original design of the SPI code,
but I'm darned if I can see a reason for it.
I left the macro definition of SPI_ERROR_CURSOR in place, so as not to
needlessly break any SPI callers that are checking for it, but that code
will never actually be returned anymore.
Tom Lane [Sun, 25 Mar 2007 19:45:14 +0000 (19:45 +0000)]
Clean up the representation of special snapshots by including a "method
pointer" in every Snapshot struct. This allows removal of the case-by-case
tests in HeapTupleSatisfiesVisibility, which should make it a bit faster
(I didn't try any performance tests though). More importantly, we are no
longer violating portable C practices by assuming that small integers are
distinct from all pointer values, and HeapTupleSatisfiesDirty no longer
has a non-reentrant API involving side-effects on a global variable.
There were a couple of places calling HeapTupleSatisfiesXXX routines
directly rather than through the HeapTupleSatisfiesVisibility macro.
Since these places had to be changed anyway, I chose to make them go
through the macro for uniformity.
Along the way I renamed HeapTupleSatisfiesSnapshot to HeapTupleSatisfiesMVCC
to emphasize that it's only used with MVCC-type snapshots. I was sorely
tempted to rename HeapTupleSatisfiesVisibility to HeapTupleSatisfiesSnapshot,
but forebore for the moment to avoid confusion and reduce the likelihood that
this patch breaks some of the pending patches. Might want to reconsider
doing that later.
Tatsuo Ishii [Sun, 25 Mar 2007 11:56:04 +0000 (11:56 +0000)]
Add new encoding EUC_JIS_2004 and SHIFT_JIS_2004,
along with new conversions among EUC_JIS_2004, SHIFT_JIS_2004 and UTF-8.
catalog version has been bump up.
Tom Lane [Fri, 23 Mar 2007 20:24:41 +0000 (20:24 +0000)]
Adjust DatumGetBool macro so that it isn't fooled by garbage in the Datum
to the left of the actual bool value. While in most cases there won't be
any, our support for old-style user-defined functions violates the C spec
to the extent of calling functions that might return char or short through
a function pointer declared to return "char *", which we then coerce to
Datum. It is not surprising that the result might contain garbage
high-order bits ... what is surprising is that we didn't see such cases
long ago. Per report from Magnus.
Tom Lane [Fri, 23 Mar 2007 19:53:52 +0000 (19:53 +0000)]
Fix plancache so that any required replanning is done with the same
search_path that was active when the plan was first made. To do this,
improve namespace.c to support a stack of "override" search path settings
(we must have a stack since nested replan events are entirely possible).
This facility replaces the "special namespace" hack formerly used by
CREATE SCHEMA, and should be able to support per-function search path
settings as well.