Tom Lane [Sat, 19 Oct 2002 20:15:09 +0000 (20:15 +0000)]
Invert logic in pg_exec_query_string() so that we set a snapshot for
all utility statement types *except* a short list, per discussion a few
days ago. Add missing SetQuerySnapshot calls in VACUUM and REINDEX,
and guard against calling REINDEX DATABASE from a function (has same
problem as VACUUM).
Tom Lane [Sat, 19 Oct 2002 19:00:47 +0000 (19:00 +0000)]
Fix rewrite code so that rules are in fact executed in order by name,
rather than being reordered according to INSTEAD attribute for
implementation convenience.
Also, increase compiled-in recursion depth limit from 10 to 100 rewrite
cycles. 10 seems pretty marginal for situations where multiple rules
exist for the same query. There was a complaint about this recently,
so I'm going to bump it up. (Perhaps we should make the limit a GUC
parameter, but that's too close to being a new feature to do in beta.)
Tom Lane [Sat, 19 Oct 2002 02:56:16 +0000 (02:56 +0000)]
Fix range-query estimation to not double-exclude NULLs, per gripe from
Ray Ontko 28-June-02. Also, fix prefix_selectivity for NAME lefthand
variables (it was bogusly assuming binary compatibility), and adjust
make_greater_string() to not call pg_mbcliplen() with invalid multibyte
data (this last per bug report that I can't find at the moment, but it
was in July '02).
Bruce Momjian [Sat, 19 Oct 2002 02:16:40 +0000 (02:16 +0000)]
It includes
-Support for mirroring tables in different Schema's
-Improved documentation for compiling with 7.1.x and 7.2.x
-Fixes a buffer overrun bug.
Bruce Momjian [Sat, 19 Oct 2002 02:09:45 +0000 (02:09 +0000)]
> Huh, I don't know where I got the idea you were (or someone else was?)
> in the position that attislocal should be reset. I'll clean everything
> up and submit the patch I had originally made.
All right, this is it. This patch merely checks if child tables have
the column. If atttypid and atttypmod are the same, the attributes'
attinhcount is incremented; else the operation is aborted. If child
tables don't have the column, recursively add it.
Bruce Momjian [Sat, 19 Oct 2002 02:08:19 +0000 (02:08 +0000)]
This patch adds some missing functions for float8 math operations,
specifically ceil(), floor(), and sign(). There may be other functions
that need to be added, but this is a start. I've included some simple
regression tests.
Tom Lane [Sat, 19 Oct 2002 00:22:14 +0000 (00:22 +0000)]
Fix psql's \copy to accept table names containing schemas, as well as
a column list. Bring its parsing of quoted names and quoted strings
somewhat up to speed --- I believe it now handles all non-error cases
the same way the backend would, but weird boundary conditions are not
necessarily done the same way.
Tom Lane [Fri, 18 Oct 2002 20:44:02 +0000 (20:44 +0000)]
Fix breakage that had crept into setlocale() usage: once again we've
been bit by the fact that the locale functions return pointers to
modifiable variables. I added some comments that might help us avoid
the mistake in future.
Tom Lane [Fri, 18 Oct 2002 20:33:57 +0000 (20:33 +0000)]
Make 'dummy' declarations in header files be 'extern int no_such_variable'
instead of 'extern int errno'; the latter is unsafe according to the
ANSI C standard, as well as in practice on some platforms.
Bruce Momjian [Wed, 16 Oct 2002 03:24:09 +0000 (03:24 +0000)]
Fix script to handle autocommit = 'off' by prepending autocommit 'on' to
the start of the psql commandline. This is better than adding BEGIN/END
because it handles multiple queries well, and allows the return code for
psql to return the proper value.
Tom Lane [Tue, 15 Oct 2002 02:24:16 +0000 (02:24 +0000)]
Fix psql to cope with autocommit off, at least during startup.
Behavior of backslash commands (especially for large objects)
may still require some thought.
Tom Lane [Mon, 14 Oct 2002 23:49:20 +0000 (23:49 +0000)]
Make SPI's execution of querystrings follow the rules agreed to for
command status at the interactive level. SPI_processed, etc are set
in the same way as the returned command status would have been set if
the same querystring were issued interactively. Per gripe from
Michael Paesold 25-Sep-02.
Tom Lane [Mon, 14 Oct 2002 22:14:35 +0000 (22:14 +0000)]
Adjust handling of command status strings in the presence of rules,
as per recent pghackers discussions. initdb forced due to change in
fields of stored Query nodes.
Tom Lane [Mon, 14 Oct 2002 16:51:30 +0000 (16:51 +0000)]
Arrange to copy relcache's trigdesc structure at the start of any
query that uses it. This ensures that triggers will be applied consistently
throughout a query even if someone commits changes to the relation's
pg_class.reltriggers field meanwhile. Per crash report from Laurette Cisneros.
While at it, simplify memory management in relcache.c, which no longer
needs the old hack to try to keep trigger info in the same place over
a relcache entry rebuild. (Should try to fix rd_att and rewrite-rule
access similarly, someday.) And make RelationBuildTriggers simpler and
more robust by making it build the trigdesc in working memory and then
CopyTriggerDesc() into cache memory.
Bruce Momjian [Mon, 14 Oct 2002 04:29:23 +0000 (04:29 +0000)]
- Link the entries in the table to the catalog heading
- Wrap them in the <database class="table"> tags, since thats what they
are (no markup rules for this, so it inherits from parent -- no style
change)
- Mention that pg_database, pg_shadow, and pg_group are global, and the
rest are local to the specific DB. (I believe this is correct).
> Works for me, though I suppose we could explain what the exceptions are
> like in general terms. Perhaps something like
>
> 'Most system catalogs are copied from the template database during
> database creation, and are thereafter database-specific. A few
> catalogs are physically shared across all databases in an installation;
> these are marked in the descriptions of the individual catalogs.'
Bruce Momjian [Mon, 14 Oct 2002 04:26:54 +0000 (04:26 +0000)]
As Niel so nicely pointed out this morning, the output of EXPLAIN
ANALYZE is not quite clear when branches of the query are never
executed. So this tiny patch fixes that.
The patch is attached and can also be found at:
http://svana.org/kleptog/pgsql/pgsql-explain.patch
Bruce Momjian [Mon, 14 Oct 2002 04:20:52 +0000 (04:20 +0000)]
I have attached two patches as per:
1) pltcl:
Add SPI_freetuptable() calls to avoid memory leaks (Me + Neil Conway)
Change sprintf()s to snprintf()s (Neil Conway)
Remove header files included elsewhere (Neil Conway)
2)plpython:
Add SPI_freetuptable() calls to avoid memory leaks
Cosemtic change to remove a compiler warning
Notes:
I have tested pltcl.c for
a) the original leak problem reported for the repeated call of spi_exec
in a TCL fragment
and
b) the subsequent report resulting from the use of spi_exec -array
in a TCL
fragment.
The plpython.c patch is exactly the same as that applied to make
revision 1.23,
the plpython_schema.sql and feature.expected sections of the patch are
also the
same as last submited, applied and subsequently reversed out. It remains
untested by me (other than via make check). However, this should be safe
provided PyString_FromString() _copies_ the given string to make a
PyObject.
Bruce Momjian [Wed, 9 Oct 2002 16:23:55 +0000 (16:23 +0000)]
Well, this patch makes Makefile for contrib/rserv use the
contrib/contrib-global.mk library and _generally_ behave like
Makefiles for other contrib modules.
Besides it fixes Perl's interpolation of $libdir variable, which
should be passed to backend instead. This patch is done against
PostgreSQL 7.3b2
Besides, I want to thank Peter Eisentraut for his very friendly and
helpful attitude and politely ask him to check whether contrib
modules actually continue to work after he implements another
major change to their build process.
Bruce Momjian [Wed, 9 Oct 2002 16:21:54 +0000 (16:21 +0000)]
> > > > and mb conversions (pg_ascii2mic and pg_mic2ascii not
> > > > found in the postmaster and not included from elsewhere)
> >
> > shared libs on AIX need to be able to resolve all symbols at linkage time.
> > Those two symbols are in backend/utils/SUBSYS.o but not in the postgres
> > executable.
>
> They are defined in backend/utils/mb/conv.c and declared in
> include/mb/pg_wchar.h. They're also linked into the
> postmaster. I don't see anything unusual.
Attached is a patch to fix the mb linking problems on AIX. As a nice side effect
it reduces the duplicate symbol warnings to linking libpq.so and libecpg.so
(all shlibs that are not postmaster loadable modules).
Please apply to current (only affects AIX).
The _LARGE_FILES problem is unfortunately still open, unless Peter
has fixed it per his recent idea.
Bruce Momjian [Wed, 9 Oct 2002 16:20:25 +0000 (16:20 +0000)]
> Alvaro Herrera <alvherre@atentus.com> writes:
> > I'm looking at pg_dump/common.c:flagInhAttrs() and suspect that it can
> > be more or less rewritten completely, and probably should to get rigth
> > all the cases mentioned in the past attisinherited discussion. Is this
> > desirable for 7.3? It can probably be hacked around and the rewrite
> > kept for 7.4, but I think it will be much simpler after the rewrite.
>
> If it's a bug then it's fair game to fix in 7.3. But keep in mind that
> pg_dump has to behave at least somewhat sanely when called against older
> servers ... will your rewrite behave reasonably if the server does not
> offer attinhcount values?
Nah. I don't think it's worth it: I had forgotten that older versions
should be supported. I just left the code as is and added a
version-specific test.
This patch allows pg_dump to dump correctly local definition of columns.
In particular,
CREATE TABLE p1 (f1 int, f2 int);
CREATE TABLE p2 (f1 int);
CREATE TABLE c () INHERITS (p1, p2);
ALTER TABLE ONLY p1 DROP COLUMN f1;
CREATE TABLE p3 (f1 int);
CREATE TABLE c2 (f1 int) INHERITS (p3);
Will be dumped as
CREATE TABLE p1 (f2 int);
CREATE TABLE p2 (f1 int);
CREATE TABLE c (f1 int) INHERITS (p1, p2);
CREATE TABLE c2 (f1 int) INHERITS (p3);
(Previous version will dump
CREATE TABLE c () INHERITS (p1, p2)
CREATE TABLE c2 () INHERITS (p3) )
Tom Lane [Tue, 8 Oct 2002 17:17:19 +0000 (17:17 +0000)]
Move responsibility for setting QuerySnapshot for utility statements
into postgres.c; make sure it happens for all cases that seem to need it.
Perhaps it would be better to explicitly exclude just a few utility
statement types from setting a snapshot?
Tom Lane [Mon, 7 Oct 2002 17:04:30 +0000 (17:04 +0000)]
Change order of operations during XLogFlush so that we try to include
in our write/flush operation any WAL entries that got queued while we
were waiting to get the WALWriteLock. This improves throughput when
transactions are small enough that several can be committed per WAL
write (ie, per disk revolution).
Tatsuo Ishii [Mon, 7 Oct 2002 05:10:02 +0000 (05:10 +0000)]
Avoid PQisBusy/PQconsumeInput busy loop in case of PQisBusy returning
false. per Tom Lane's suggestion. See:
Subject: Suggested change to pgbench
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Tatsuo Ishii <t-ishii@sra.co.jp> Cc: pgsql-patches@postgreSQL.org
Date: Sun, 06 Oct 2002 12:37:27 -0400
Tom Lane [Fri, 4 Oct 2002 22:08:44 +0000 (22:08 +0000)]
Require superuser privilege to create a binary-compatible cast, per
discussion some weeks ago. Also, add a check that two types to be
binary-equivalenced match as to typlen, typbyval, and typalign; if
they don't then it's surely a mistake to equivalence them.
Tom Lane [Fri, 4 Oct 2002 17:19:55 +0000 (17:19 +0000)]
Tweak a few of the most heavily used function call points to zero out
just the significant fields of FunctionCallInfoData, rather than MemSet'ing
the whole struct to zero. Unused positions in the arg[] array will
thereby contain garbage rather than zeroes. This buys back some of the
performance hit from increasing FUNC_MAX_ARGS. Also tweak tuplesort.c
code for more speed by marking some routines 'inline'. All together
these changes speed up simple sorts, like count(distinct int4column),
by about 25% on a P4 running RH Linux 7.2.
Tom Lane [Thu, 3 Oct 2002 21:06:23 +0000 (21:06 +0000)]
Hack to make it possible to load CREATE CONSTRAINT TRIGGER commands that
are missing the FROM clause (due to a long-ago pg_dump bug). Patch by
Stephan Szabo, minor tweaking by Tom Lane.