Tom Lane [Fri, 27 Sep 2002 15:04:08 +0000 (15:04 +0000)]
Don't mess with HEAP_XMAX_INVALID in heaptuple.c routines; there is
no reason to worry about the tuple commit status bits until the tuple
is inserted in a relation by heapam.c. Also, improve comments for
heap_addheader().
Tom Lane [Thu, 26 Sep 2002 22:58:34 +0000 (22:58 +0000)]
Ensure that before truncating CLOG, we force a checkpoint even if no
recent WAL activity has occurred. Without this, it's possible that a
later crash might leave tuples on disk with un-updated commit status
bits.
Tom Lane [Thu, 26 Sep 2002 22:46:29 +0000 (22:46 +0000)]
Fix problems with loss of tuple commit status bits during WAL redo of
VACUUM FULL tuple moves. Store full-width t_infomask in WAL, rather
than storing low 8 bits and expecting to be able to reconstruct upper
bits. While at it, remove redundant t_oid field from WAL headers
(the OID, if present, is now recorded in the data portion of the tuple).
WAL version number bumped --- this does not force an initdb, you can
instead run pg_resetxlog after a clean shutdown of the old postmaster.
Bruce Momjian [Thu, 26 Sep 2002 05:24:30 +0000 (05:24 +0000)]
Attached is a patch to contrib/dbmirror that fixes a bug that was
causing the postmaster to crash when the trigger was running on a table
without a primary key.
I've also updated the docs to explicitly say that tables need primary
keys.
Bruce Momjian [Thu, 26 Sep 2002 05:23:26 +0000 (05:23 +0000)]
I have attached the pltcl patch again, just in case. For the sake of clarity
let's say this patch superscedes the previous one.
I have also attached a patch addressing the similar memory leak problem in
plpython. This includes a slight adjustment of the tests in the source
directory. The patch also includes a cosmetic change to remove a compiler
warning although I think the change makes the code look worse though.
BTW, by my reckoning the memory leak would occur with prepared plans and
without. If that is not the case then I've been barking up the wrong tree.
Tom Lane [Wed, 25 Sep 2002 20:31:40 +0000 (20:31 +0000)]
Remove ShutdownBufferPoolAccess exit callback, and do the work in
ProcKill instead, where we still have a PGPROC with which to wait on
LWLocks. This fixes 'can't wait without a PROC structure' failures
occasionally seen during backend shutdown (I'm surprised they weren't
more frequent, actually). Add an Assert() to LWLockAcquire to help
catch any similar mistakes in future. Fix failure to update MyProcPid
for standalone backends and pgstat processes.
Tom Lane [Tue, 24 Sep 2002 23:14:25 +0000 (23:14 +0000)]
Change default privileges for languages and functions to be PUBLIC USAGE
and PUBLIC EXECUTE, respectively. Per discussion about easing updates
from prior versions.
Tom Lane [Tue, 24 Sep 2002 18:38:23 +0000 (18:38 +0000)]
Add missing correction of sublevelsup when pulling up a subquery.
Fixes problem with cases like
SELECT * FROM foo t WHERE NOT EXISTS (SELECT remoteid FROM
(SELECT f1 as remoteid FROM foo WHERE f1 = t.f1) AS t1)
Tom Lane [Mon, 23 Sep 2002 22:57:44 +0000 (22:57 +0000)]
Further thinking about heap_mark4update: in HeapTupleSelfUpdated case,
executor should not return the tuple as successfully marked, because in
fact it's been deleted. Not clear that this case has ever been seen
in practice (I think you'd have to write a SELECT FOR UPDATE that calls
a function that deletes some row the SELECT will visit later...) but we
should be consistent. Also add comments to several other places that
got it right but didn't explain what they were doing.
Tom Lane [Mon, 23 Sep 2002 20:43:41 +0000 (20:43 +0000)]
Disallow VACUUM, ANALYZE, TRUNCATE on temp tables belonging to other
backends. Given that temp tables now store data locally in the local
buffer manager, these things are not going to work safely.
Bruce Momjian [Mon, 23 Sep 2002 01:51:02 +0000 (01:51 +0000)]
The valid return value should be MODIFY instead of MODIFIED.
The error message said so :-)
In 25.3. Using PL/Python
If the trigger "when" is BEFORE, you may return None or "OK"
from the Python function to indicate the tuple is unmodified, "SKIP"
to abort the event, or "MODIFIED" to indicate you've modified the tuple.
should read
If the trigger "when" is BEFORE, you may return None or "OK"
from the Python function to indicate the tuple is unmodified, "SKIP"
to abort the event, or "MODIFY" to indicate you've modified the tuple.
Bruce Momjian [Mon, 23 Sep 2002 01:43:23 +0000 (01:43 +0000)]
Obviously noone has ever tested the doubling of availiable result ids
up to
reaching the hard limit. After opening 16(=current REST_START value)
results via pg_exec, the next pg_exec tries to find an empty slot
forever :-( . In PgSetResultId file pgtclId.c in the for loop there
has to be done a break, if res_max ist reached. The piece of code
should look like
if (resid == connid->res_max)
{
resid = 0;
break; /* the break as to be added */
}
now everything works (double available results after reaching
RES_START up to reaching RES_HARD_MAX)
Tom Lane [Mon, 23 Sep 2002 00:42:48 +0000 (00:42 +0000)]
Get rid of bogus use of heap_mark4update in reindex operations (cf.
recent bug report). Fix processing of nailed-in-cache indexes;
it appears that REINDEX DATABASE has been broken for months :-(.
Tom Lane [Sun, 22 Sep 2002 23:03:58 +0000 (23:03 +0000)]
In UpdateStats(), don't bother to update the pg_class row if it already
contains the correct statistics. This is a partial solution for the
problem of allowing concurrent CREATE INDEX commands: unless they commit
at nearly the same instant, the second one will see the first one's
pg_class updates as committed, and won't try to update again, thus
avoiding the 'tuple concurrently updated' failure.
Tom Lane [Sun, 22 Sep 2002 20:56:28 +0000 (20:56 +0000)]
RelationClearRelation must reset rd_targblock and update rd_nblocks
even when dealing with a nailed-in-cache relation; otherwise, following
VACUUM truncation of a system catalog, other backends might have
unreasonably large values of these fields.
Tom Lane [Sun, 22 Sep 2002 20:08:51 +0000 (20:08 +0000)]
I haven't seen any objections, so here is a patch. It removes "#define
YYERROR_VERBOSE" from contrib/cube and contrib/seg, and adjusts the expected
output accordingly. Hopefully this will consistently pass across multiple
bison versions.
Tom Lane [Sun, 22 Sep 2002 19:42:52 +0000 (19:42 +0000)]
Replace pg_attribute.attisinherited with attislocal and attinhcount
columns, to allow more correct behavior in multiple-inheritance cases.
Patch by Alvaro Herrera, review by Tom Lane.
Tom Lane [Sun, 22 Sep 2002 17:27:25 +0000 (17:27 +0000)]
Bring SIMILAR TO and SUBSTRING into some semblance of conformance with
the SQL99 standard. (I'm not sure that the character-class features are
quite right, but that can be fixed later.) Document SQL99 and POSIX
regexps as being different features; provide variants of SUBSTRING for
each.
Tom Lane [Sun, 22 Sep 2002 00:37:09 +0000 (00:37 +0000)]
Move most of the error checking for foreign-key constraints out of
parse analysis and into the execution code (in tablecmds.c). This
eliminates a lot of unreasonably complex code that needed to have two
or more execution paths in case it was dealing with a not-yet-created
table column vs. an already-existing one. The execution code is always
dealing with already-created tables and so needs only one case. This
also eliminates some potential race conditions (the table wasn't locked
between parse analysis and execution), makes it easy to fix the gripe
about wrong referenced-column names generating a misleading error message,
and lets us easily add a dependency from the foreign-key constraint to
the unique index that it requires the referenced table to have. (Cf.
complaint from Kris Jurka 12-Sep-2002 on pgsql-bugs.)
Also, third try at building a deletion mechanism that is not sensitive
to the order in which pg_depend entries are visited. Adding the above-
mentioned dependency exposed the folly of what dependency.c had been
doing: it failed for cases where B depends on C while both auto-depend
on A. Dropping A should succeed in this case, but was failing if C
happened to be visited before B. It appears the only solution is two
separate walks over the dependency tree.
Tom Lane [Sat, 21 Sep 2002 18:39:26 +0000 (18:39 +0000)]
Provide an upgrade strategy for dump files containing functions declared
with OPAQUE. CREATE LANGUAGE, CREATE TRIGGER, and CREATE TYPE will all
accept references to functions declared with OPAQUE --- but they will
issue a NOTICE, and will modify the function entries in pg_proc to have
the preferred type-safe argument or result types instead of OPAQUE.
Per recent pghackers discussions.
Tom Lane [Fri, 20 Sep 2002 19:56:01 +0000 (19:56 +0000)]
Performance improvement for MultiRecordFreeSpace on large relations ---
avoid O(N^2) behavior. Problem noted and fixed by Stephen Marshall <smarshall@wsicorp.com>,
with some help from Tom Lane.
Tom Lane [Fri, 20 Sep 2002 16:56:02 +0000 (16:56 +0000)]
Code review for check-domain-constraints-in-COPY patch. Do correct thing
when default expression for a domain is being used. Avoid repetitive
catalog lookups.
Bruce Momjian [Fri, 20 Sep 2002 03:55:40 +0000 (03:55 +0000)]
Tom Lane wrote:
> I see in your recent bytea-LIKE patch
>
> if (datatype != BYTEAOID && pg_database_encoding_max_length()
> 1)
> len = pg_mbcliplen((const unsigned char *) workstr, len,
len - 1);
> else
> len -= -1;
>
> Surely there's one too many minus signs in that last?
Bruce Momjian [Fri, 20 Sep 2002 03:54:57 +0000 (03:54 +0000)]
This patch cleans up some of the code in src/util/adt/formatting.c,
fixes a few minor bugs (typos, potential buffer overruns, etc.), and
fixes some spelling/grammar mistakes.
Bruce Momjian [Fri, 20 Sep 2002 03:47:22 +0000 (03:47 +0000)]
I have included fixes to declare some floating point constants as double
instead of int, change the calculation method to use the haversine
formula
which is more accurrate for short distances, added a grant to public for
geo_distance and added a regression test.
I will resubmit the earth distance stuff based on cube after 7.3 is
released.
Bruce Momjian [Fri, 20 Sep 2002 03:45:08 +0000 (03:45 +0000)]
> I'm not sure why NDirectFileRead/NDirectFileWrite are unused at the
> moment, but they used to be used; I think the correct response is to
> put back the missing counter increments, not rip out the counters.
Ok, fair enough. It's worth noting that they've been broken for a
while -- for example, the HashJoin counter increments were broken when
you comitted r1.20 of executor/nodeHashJoin.c in May of '99.
I've attached a revised patch that doesn't remove the counters (but
doesn't increment them either: I'm not sure of all the places where
the counter should be incremented).
Bruce Momjian [Fri, 20 Sep 2002 03:39:15 +0000 (03:39 +0000)]
This patch corrects a minor mis-statement in the CREATE DOMAIN docs,
pointed out by Chris KL on -hackers a little while ago -- NOT NULL !=
CHECK xxx IS NOT NULL.
Tom Lane [Thu, 19 Sep 2002 23:40:56 +0000 (23:40 +0000)]
Tweak heap.c to refuse attempts to create table columns of standalone
composite types. Add a couple more lsyscache.c routines to support this,
and make use of them in some other places that were doing lookups the
hard way.
Tom Lane [Thu, 19 Sep 2002 22:48:34 +0000 (22:48 +0000)]
Department of second thoughts: suppressing implicit casts everywhere in
ruleutils display is not such a great idea. For arguments of functions
and operators I think we'd better keep the historical behavior of showing
such casts explicitly, to ensure that the function/operator is reparsed
the same way when the rule is reloaded. This also makes the output of
EXPLAIN less obscurantist about exactly what's happening.
Tom Lane [Wed, 18 Sep 2002 21:35:25 +0000 (21:35 +0000)]
Extend pg_cast castimplicit column to a three-way value; this allows us
to be flexible about assignment casts without introducing ambiguity in
operator/function resolution. Introduce a well-defined promotion hierarchy
for numeric datatypes (int2->int4->int8->numeric->float4->float8).
Change make_const to initially label numeric literals as int4, int8, or
numeric (never float8 anymore).
Explicitly mark Func and RelabelType nodes to indicate whether they came
from a function call, explicit cast, or implicit cast; use this to do
reverse-listing more accurately and without so many heuristics.
Explicit casts to char, varchar, bit, varbit will truncate or pad without
raising an error (the pre-7.2 behavior), while assigning to a column without
any explicit cast will still raise an error for wrong-length data like 7.3.
This more nearly follows the SQL spec than 7.2 behavior (we should be
reporting a 'completion condition' in the explicit-cast cases, but we have
no mechanism for that, so just do silent truncation).
Fix some problems with enforcement of typmod for array elements;
it didn't work at all in 'UPDATE ... SET array[n] = foo', for example.
Provide a generalized array_length_coerce() function to replace the
specialized per-array-type functions that used to be needed (and were
missing for NUMERIC as well as all the datetime types).
Add missing conversions int8<->float4, text<->numeric, oid<->int8.
initdb forced.
Tom Lane [Mon, 16 Sep 2002 01:24:41 +0000 (01:24 +0000)]
Fix for rare race-condition-like failure: if a backend receives SIGUSR2
(notify/SI-overrun interrupt) while it is in process of doing proc_exit,
it is possible for Async_NotifyHandler() to try to start a transaction
when one is already running. This leads to Asserts() or worse. I think
it may only be possible to occur when frontend synchronization is lost
(ie, the elog(FATAL) in SocketBackend() fires), but that is a standard
occurrence after error during COPY. In any case, I have seen this
failure occur during regression tests, so it is definitely possible.