OSDN Git Service
Michael Meskes [Thu, 26 Nov 2009 15:40:43 +0000 (15:40 +0000)]
Forgot to add dynamic cursors to Changelog.
Michael Meskes [Thu, 26 Nov 2009 15:39:26 +0000 (15:39 +0000)]
Synced addon rules after some renamings.
Michael Meskes [Thu, 26 Nov 2009 15:06:47 +0000 (15:06 +0000)]
Added dynamic cursor names to ecpg. Almost the whole patch was done by
Boszormenyi Zoltan, with only a minor tweak or two from me.
Tom Lane [Wed, 25 Nov 2009 20:26:31 +0000 (20:26 +0000)]
Simplify psql's new linestyle behavior to default to linestyle=ascii all
the time, rather than hoping we can tell whether the terminal supports
UTF8 characters. Per discussion.
Peter Eisentraut [Tue, 24 Nov 2009 19:21:15 +0000 (19:21 +0000)]
Fix syntax in extract() examples
Author: Erik Rijkers <er@xs4all.nl>
Michael Meskes [Tue, 24 Nov 2009 16:30:31 +0000 (16:30 +0000)]
Made function better readable.
Peter Eisentraut [Mon, 23 Nov 2009 21:41:20 +0000 (21:41 +0000)]
Add PG_MODULE_MAGIC and some missing include files to examples
Author: Euler Taveira de Oliveira <euler@timbira.com>
Tom Lane [Mon, 23 Nov 2009 16:02:24 +0000 (16:02 +0000)]
Use diff's -w switch only on Windows, to avoid problems with inconsistent
newline representations. Per buildfarm results and subsequent discussion.
Sync up a couple of other places that had their own policies.
Magnus Hagander [Mon, 23 Nov 2009 12:54:37 +0000 (12:54 +0000)]
Add missing library to standalone libpq build on Win32.
Hiroshi Saito
Heikki Linnakangas [Mon, 23 Nov 2009 09:58:36 +0000 (09:58 +0000)]
Fix an old bug in multixact and two-phase commit. Prepared transactions can
be part of multixacts, so allocate a slot for each prepared transaction in
the "oldest member" array in multixact.c. On PREPARE TRANSACTION, transfer
the oldest member value from the current backends slot to the prepared xact
slot. Also save and recover the value from the 2pc state file.
The symptom of the bug was that after a transaction prepared, a shared lock
still held by the prepared transaction was sometimes ignored by other
transactions.
Fix back to 8.1, where both 2PC and multixact were introduced.
Tom Lane [Sun, 22 Nov 2009 22:06:30 +0000 (22:06 +0000)]
Assorted wordsmithing on the documentation of \pset --- try to make it
a bit more consistent and less obviously written by different people at
different times.
Tom Lane [Sun, 22 Nov 2009 20:58:51 +0000 (20:58 +0000)]
Adjust expected-results spacing per buildfarm results.
Tom Lane [Sun, 22 Nov 2009 17:54:23 +0000 (17:54 +0000)]
Remove -w (--ignore-all-space) option from pg_regress's diff calls.
We have used -w for a long time as a means of reducing the reported diff
volume when one element of a result table isn't of the expected width.
However, most of the time the results just pass anyway, so this isn't as
important as it once was. Meanwhile, the risk of missing potentially
significant deviations has gone up, particularly with psql's ability to
report error cursor positions. So, let's switch over to space-sensitive
comparisons. Per my proposal of yesterday.
(All the expected files that I can test here seem to be ready for this
already, but we'll see what the buildfarm thinks about others.)
Heikki Linnakangas [Sun, 22 Nov 2009 14:54:31 +0000 (14:54 +0000)]
Remove superfluous curly brace, fixing compilation with OPTIMIZER_DEBUG.
Jan Urbanski
Tom Lane [Sun, 22 Nov 2009 06:01:09 +0000 (06:01 +0000)]
Ooops, forgot to check the libxml cases for psql output wrapping change.
Tom Lane [Sun, 22 Nov 2009 05:20:41 +0000 (05:20 +0000)]
Improve psql's tabular display of wrapped-around data by inserting markers
in the formerly-always-blank columns just to left and right of the data.
Different marking is used for a line break caused by a newline in the data
than for a straight wraparound. A newline break is signaled by a "+" in the
right margin column in ASCII mode, or a carriage return arrow in UNICODE mode.
Wraparound is signaled by a dot in the right margin as well as the following
left margin in ASCII mode, or an ellipsis symbol in the same places in UNICODE
mode. "\pset linestyle old-ascii" is added to make the previous behavior
available if anyone really wants it.
In passing, this commit also cleans up a few regression test files that
had unintended spacing differences from the current actual output.
Roger Leigh, reviewed by Gabrielle Roth and other members of PDXPUG.
Peter Eisentraut [Sat, 21 Nov 2009 23:59:12 +0000 (23:59 +0000)]
Ignore UTF-8-encoded Unicode byte-order mark at the beginning of a file if
the client encoding is UTF-8.
a limited version of a patch proposed by Itagaki Takahiro
Tom Lane [Sat, 21 Nov 2009 05:44:05 +0000 (05:44 +0000)]
Refactor ecpg grammar so that it uses the core grammar's unreserved_keyword
list, minus a few specific words that have to be treated specially. This
replaces a hard-wired list of keywords that would have needed manual
maintenance, and was not getting it. The 8.4 coding was already missing
these words, causing ecpg to incorrectly treat them as reserved words:
CALLED, CATALOG, DEFINER, ENUM, FOLLOWING, INVOKER, OPTIONS, PARTITION,
PRECEDING, RANGE, SECURITY, SERVER, UNBOUNDED, WRAPPER. In HEAD we were
additionally missing COMMENTS, FUNCTIONS, SEQUENCES, TABLES.
Per gripe from Bosco Rama.
Tom Lane [Fri, 20 Nov 2009 20:38:12 +0000 (20:38 +0000)]
Add a WHEN clause to CREATE TRIGGER, allowing a boolean expression to be
checked to determine whether the trigger should be fired.
For BEFORE triggers this is mostly a matter of spec compliance; but for AFTER
triggers it can provide a noticeable performance improvement, since queuing of
a deferred trigger event and re-fetching of the row(s) at end of statement can
be short-circuited if the trigger does not need to be fired.
Takahiro Itagaki, reviewed by KaiGai Kohei.
Peter Eisentraut [Thu, 19 Nov 2009 22:05:48 +0000 (22:05 +0000)]
Typo: dump -> restore
fixed in 8.4 and 8.5
Author: Guillaume Lelarge <guillaume@lelarge.info>
Tom Lane [Thu, 19 Nov 2009 02:45:33 +0000 (02:45 +0000)]
Fix memory leak in syslogger: logfile_rotate() would leak a copy of the
output filename if CSV logging was enabled and only one of the two possible
output files got rotated during a particular call (which would, in fact,
typically be the case during a size-based rotation). This would amount to
about MAXPGPATH (1KB) per rotation, and it's been there since the CSV
code was put in, so it's surprising that nobody noticed it before.
Per bug #5196 from Thomas Poindessous.
Tom Lane [Wed, 18 Nov 2009 21:57:56 +0000 (21:57 +0000)]
Add a hook to CREATE/ALTER ROLE to allow an external module to check the
strength of database passwords, and create a sample implementation of
such a hook as a new contrib module "passwordcheck".
Laurenz Albe, reviewed by Takahiro Itagaki
Tom Lane [Mon, 16 Nov 2009 21:32:07 +0000 (21:32 +0000)]
Provide a parenthesized-options syntax for VACUUM, analogous to that recently
adopted for EXPLAIN. This will allow additional options to be implemented
in future without having to make them fully-reserved keywords. The old syntax
remains available for existing options, however.
Itagaki Takahiro
Tom Lane [Mon, 16 Nov 2009 18:04:40 +0000 (18:04 +0000)]
While doing the final setrefs.c pass over a plan tree, try to match up
non-Var sort/group expressions using ressortgroupref labels instead of
depending entirely on equal()-ity of the upper node's tlist expressions
to the lower node's. This avoids emitting the wrong outputs in cases
where there are textually identical volatile sort/group expressions,
as for example
select distinct random(),random() from generate_series(1,10);
Per report from Andrew Gierth.
Backpatch to 8.4. Arguably this is wrong all the way back, but the only known
case where there's an observable problem is when using hash aggregation to
implement DISTINCT, which is new as of 8.4. So for the moment I'll refrain
from backpatching further.
Peter Eisentraut [Sun, 15 Nov 2009 13:57:01 +0000 (13:57 +0000)]
Make text search parser accept underscores in XML attributes (bug #5075)
Tom Lane [Sun, 15 Nov 2009 02:45:35 +0000 (02:45 +0000)]
Improve planning of Materialize nodes inserted atop the inner input of a
mergejoin to shield it from doing mark/restore and refetches. Put an explicit
flag in MergePath so we can centralize the logic that knows about this,
and add costing logic that considers using Materialize even when it's not
forced by the previously-existing considerations. This is in response to
a discussion back in August that suggested that materializing an inner
indexscan can be helpful when the refetch percentage is high enough.
Tom Lane [Sat, 14 Nov 2009 18:24:32 +0000 (18:24 +0000)]
Make unaccent's install/uninstall scripts look more like all the others.
Set search_path explicitly, don't use IF EXISTS, etc.
Magnus Hagander [Sat, 14 Nov 2009 15:39:36 +0000 (15:39 +0000)]
Add inheritable ACE when creating a restricted token for execution on
Win32.
Also refactor the code around it to be more clear.
Jesse Morris
Tom Lane [Fri, 13 Nov 2009 23:49:23 +0000 (23:49 +0000)]
Clean up a couple of bizarre code formatting choices in recent CREATE LIKE patch.
Tom Lane [Fri, 13 Nov 2009 23:44:19 +0000 (23:44 +0000)]
Avoid assuming that enum CreateStmtLikeOption is unsigned. Zdenek Kotala
Tom Lane [Fri, 13 Nov 2009 22:43:42 +0000 (22:43 +0000)]
Add control knobs for plpgsql's variable resolution behavior, and make the
default be "throw error on conflict", as per discussions. The GUC variable
is plpgsql.variable_conflict, with values "error", "use_variable",
"use_column". The behavior can also be specified per-function by inserting
one of
#variable_conflict error
#variable_conflict use_variable
#variable_conflict use_column
at the start of the function body.
The 8.5 release notes will need to mention using "use_variable" to retain
backward-compatible behavior, although we should encourage people to migrate
to the much less mistake-prone "error" setting.
Update the plpgsql documentation to match this and other recent changes.
Heikki Linnakangas [Fri, 13 Nov 2009 19:48:20 +0000 (19:48 +0000)]
A better fix for the "ARRAY[...]::domain" problem. The previous patch worked,
but the transformed ArrayExpr claimed to have a return type of "domain",
even though the domain constraint was only checked by the enclosing
CoerceToDomain node. With this fix, the ArrayExpr is correctly labeled with
the base type of the domain. Per gripe by Tom Lane.
Heikki Linnakangas [Fri, 13 Nov 2009 16:09:10 +0000 (16:09 +0000)]
When you do "ARRAY[...]::domain", where domain is a domain over an array type,
we need to check domain constraints. We used to do it correctly, but 8.4
introduced a separate code path for the "ARRAY[]::arraytype" case to infer
the type of an empty ARRAY construct from the cast target, and forgot to take
domains into account.
Per report from Florian G. Pflug.
Teodor Sigaev [Fri, 13 Nov 2009 11:17:04 +0000 (11:17 +0000)]
Fix multicolumn GIN's wrong results with fastupdate enabled.
User-defined consistent functions believes the check array
contains at least one true element which was not a true for
scanning pending list.
Per report from Yury Don <yura@vpcit.ru>
Tom Lane [Thu, 12 Nov 2009 18:20:23 +0000 (18:20 +0000)]
The recent patch to log changes in postgresql.conf settings dumped core
if the initial value of a string variable was NULL, which is entirely
possible. Noted while experimenting with custom_variable_classes.
Tom Lane [Thu, 12 Nov 2009 03:37:15 +0000 (03:37 +0000)]
Check for C/POSIX before assuming that nl_langinfo or win32_langinfo
will work. Per buildfarm results.
Tom Lane [Thu, 12 Nov 2009 02:46:16 +0000 (02:46 +0000)]
Make initdb behave sanely when the selected locale has codeset "US-ASCII".
Per discussion, this should result in defaulting to SQL_ASCII encoding.
The original coding could not support that because it conflated selection
of SQL_ASCII encoding with not being able to determine the encoding.
Adjust pg_get_encoding_from_locale()'s API to distinguish these cases,
and fix callers appropriately. Only initdb actually changes behavior,
since the other callers were perfectly content to consider these cases
equivalent.
Per bug #5178 from Boh Yap. Not going to bother back-patching, since
no one has complained before and there's an easy workaround (namely,
specify the encoding you want).
Tom Lane [Thu, 12 Nov 2009 01:13:12 +0000 (01:13 +0000)]
Remove pg_parse_string_token() --- not needed anymore.
Tom Lane [Thu, 12 Nov 2009 00:13:00 +0000 (00:13 +0000)]
Remove plpgsql's separate lexer (finally!), in favor of using the core lexer
directly. This was a lot of trouble, but should be worth it in terms of
not having to keep the plpgsql lexer in step with core anymore. In addition
the handling of keywords is significantly better-structured, allowing us to
de-reserve a number of words that plpgsql formerly treated as reserved.
Peter Eisentraut [Wed, 11 Nov 2009 21:07:41 +0000 (21:07 +0000)]
In psql \du, separate the role attributes by comma instead of newline,
for an arguably more pleasant display.
Alvaro Herrera [Wed, 11 Nov 2009 20:31:26 +0000 (20:31 +0000)]
Change "name" nonterminal in cursor-related productions to cursor_name.
This is a preparatory patch for allowing a dynamic cursor name be used in the
ECPG grammar.
Author: Zoltan Boszormenyi
Alvaro Herrera [Wed, 11 Nov 2009 20:07:57 +0000 (20:07 +0000)]
Document the previous FETCH and MOVE changes.
Alvaro Herrera [Wed, 11 Nov 2009 19:25:42 +0000 (19:25 +0000)]
Support optional FROM/IN in FETCH and MOVE
The main motivation for this is that it's required for Informix compatibility
in ECPG.
This patch makes the ECPG and core grammars a bit closer to one another for
these productions.
Author: Zoltan Boszormenyi
Tom Lane [Tue, 10 Nov 2009 23:12:13 +0000 (23:12 +0000)]
Do not build psql's flex module on its own, but instead include it in
mainloop.c. This ensures that postgres_fe.h is read before including
any system headers, which is necessary to avoid problems on some platforms
where we make nondefault selections of feature macros for stdio.h or
other headers. We have had this policy for flex modules in the backend
for many years, but for some reason it was not applied to psql.
Per trouble report from Alexandra Roy and diagnosis by Albe Laurenz.
Tom Lane [Tue, 10 Nov 2009 18:53:38 +0000 (18:53 +0000)]
Revert the temporary patch to work around Snow Leopard readdir() bug.
Apple has fixed that bug in 10.6.2, and we should encourage users to
update to that version rather than trusting this cosmetic patch.
As was recently noted by Stephen Tyler, this patch was only masking
the problem in the context of DROP TABLESPACE, but the failure could
occur in other places such as pg_xlog cleanup.
Bruce Momjian [Tue, 10 Nov 2009 18:41:24 +0000 (18:41 +0000)]
interval_abs():
Add C comment about why there is no interval_abs(): it is unclear what
value to return:
http://archives.postgresql.org/pgsql-general/2009-10/msg01031.php
http://archives.postgresql.org/pgsql-general/2009-11/msg00041.php
Alvaro Herrera [Tue, 10 Nov 2009 18:00:06 +0000 (18:00 +0000)]
Fix longstanding problems in VACUUM caused by untimely interruptions
In VACUUM FULL, an interrupt after the initial transaction has been recorded
as committed can cause postmaster to restart with the following error message:
PANIC: cannot abort transaction NNNN, it was already committed
This problem has been reported many times.
In lazy VACUUM, an interrupt after the table has been truncated by
lazy_truncate_heap causes other backends' relcache to still point to the
removed pages; this can cause future INSERT and UPDATE queries to error out
with the following error message:
could not read block XX of relation 1663/NNN/MMMM: read only 0 of 8192 bytes
The window to this race condition is extremely narrow, but it has been seen in
the wild involving a cancelled autovacuum process.
The solution for both problems is to inhibit interrupts in both operations
until after the respective transactions have been committed. It's not a
complete solution, because the transaction could theoretically be aborted by
some other error, but at least fixes the most common causes of both problems.
Bruce Momjian [Tue, 10 Nov 2009 14:22:45 +0000 (14:22 +0000)]
DIAGNOSTICS/FOUND wording
Update wording of GET DIAGNOSTICS/FOUND, per David Fetter.
Tom Lane [Tue, 10 Nov 2009 02:13:13 +0000 (02:13 +0000)]
More incremental refactoring in plpgsql: get rid of gram.y dependencies on
yytext. This is a necessary change if we're going to have a lexer interface
layer that does lookahead, since yytext won't necessarily be in step with
what the grammar thinks is the current token. yylval and yylloc should
be the only side-variables that we need to manage when doing lookahead.
Bruce Momjian [Tue, 10 Nov 2009 02:09:54 +0000 (02:09 +0000)]
PL/pgSQL FOUND
Document that GET DIAGNOSTICS is affected by EXECUTE, while FOUND is
not.
Tom Lane [Mon, 9 Nov 2009 18:38:48 +0000 (18:38 +0000)]
Re-refactor the core scanner's API, in order to get out from under the problem
of different parsers having different YYSTYPE unions that they want to use
with it. I defined a new union core_YYSTYPE that is just the (very short)
list of semantic values returned by the core scanner. I had originally
worried that this would require an extra interface layer, but actually we can
have parser.c's base_yylex (formerly filtered_base_yylex) take care of that at
no extra cost. Names associated with the core scanner are now "core_yy_foo",
with "base_yy_foo" being used in the core Bison parser and the parser.c
interface layer.
This solves the last serious stumbling block to eliminating plpgsql's separate
lexer. One restriction that will still be present is that plpgsql and the
core will have to agree on the token numbers assigned to tokens that can be
returned by the core lexer. Since Bison doesn't seem willing to accept
external assignments of those numbers, we'll have to live with decreeing that
core and plpgsql grammars declare these tokens first and in the same order.
Tom Lane [Mon, 9 Nov 2009 02:36:59 +0000 (02:36 +0000)]
Fix WHERE CURRENT OF to work as designed within plpgsql. The argument
can be the name of a plpgsql cursor variable, which formerly was converted
to $N before the core parser saw it, but that's no longer the case.
Deal with plain name references to plpgsql variables, and add a regression
test case that exposes the failure.
Tom Lane [Mon, 9 Nov 2009 00:26:55 +0000 (00:26 +0000)]
Modernize plpgsql's handling of parse locations, making it look a lot more
like the core parser's code. In particular, track locations at the character
rather than line level during parsing, allowing many more parse-time error
conditions to be reported with precise error pointers rather than just
"near line N".
Also, exploit the fact that we no longer need to substitute $N for variable
references by making extracted SQL queries and expressions be exact copies
of subranges of the function text, rather than having random whitespace
changes within them. This makes it possible to directly map parse error
positions from the core parser onto positions in the function text, which
lets us report them without the previous kluge of showing the intermediate
internal-query form. (Later it might be good to do that for core
parse-analysis errors too, but this patch is just touching plpgsql's
lexer/parser, not what happens at runtime.)
In passing, make plpgsql's lexer use palloc not malloc.
These changes make plpgsql's parse-time error reports noticeably nicer
(as illustrated by the regression test changes), and will also simplify
the planned removal of plpgsql's separate lexer by reducing the impedance
mismatch between what it does and what the core lexer does.
Tom Lane [Sat, 7 Nov 2009 17:21:34 +0000 (17:21 +0000)]
Remove ancient text file containing plpgsql installation instructions.
This was long ago superseded by the standard build process and main
SGML documentation.
Tom Lane [Sat, 7 Nov 2009 00:52:26 +0000 (00:52 +0000)]
Rearrange plpgsql parsing to simplify and speed it up a bit.
* Pull the responsibility for %TYPE and %ROWTYPE out of the scanner,
letting read_datatype manage it instead.
* Avoid unnecessary scanner-driven lookups of plpgsql variables in
places where it's not needed, which is actually most of the time;
we do not need it in DECLARE sections nor in text that is a SQL
query or expression.
* Rationalize the set of token types returned by the scanner:
distinguishing T_SCALAR, T_RECORD, T_ROW seems to complicate the grammar
in more places than it simplifies it, so merge these into one
token type T_DATUM; but split T_ERROR into T_DBLWORD and T_TRIPWORD
for clarity and simplicity of later processing.
Some of this will need to be revisited again when we try to make
plpgsql use the core scanner, but this patch gets some of the bigger
stumbling blocks out of the way.
Andrew Dunstan [Fri, 6 Nov 2009 21:57:57 +0000 (21:57 +0000)]
Keep track of language's trusted flag in InlineCodeBlock. Needed to support DO blocks for languages that have both trusted and untrusted variants.
Tom Lane [Fri, 6 Nov 2009 18:37:55 +0000 (18:37 +0000)]
Change plpgsql from using textual substitution to insert variable references
into SQL expressions, to using the newly added parser callback hooks.
This allows us to do the substitutions in a more semantically-aware way:
a variable reference will only be recognized where it can validly go,
ie, a place where a column value or parameter would be legal, instead of
the former behavior that would replace any textual match including
table names and column aliases (leading to syntax errors later on).
A release-note-worthy fine point is that plpgsql variable names that match
fully-reserved words will now need to be quoted.
This commit preserves the former behavior that variable references take
precedence over any possible match to a column name. The infrastructure
is in place to support the reverse precedence or throwing an error on
ambiguity, but those behaviors aren't accessible yet.
Most of the code changes here are associated with making the namespace
data structure persist so that it can be consulted at runtime, instead
of throwing it away at the end of initial function parsing.
The plpgsql scanner is still doing name lookups, but that behavior is
now irrelevant for SQL expressions. A future commit will deal with
removing unnecessary lookups.
Tom Lane [Thu, 5 Nov 2009 23:24:27 +0000 (23:24 +0000)]
Don't treat NEW and OLD as reserved words anymore. For the purposes of rules
it works just as well to have them be ordinary identifiers, and this gets rid
of a number of ugly special cases. Plus we aren't interfering with non-rule
usage of these names.
catversion bump because the names change internally in stored rules.
Peter Eisentraut [Thu, 5 Nov 2009 20:13:06 +0000 (20:13 +0000)]
reenable -> re-enable
Pointed out by Debian's lintian.
Tom Lane [Thu, 5 Nov 2009 16:58:36 +0000 (16:58 +0000)]
Remove plpgsql's RENAME declaration, which has bizarre and mostly nonfunctional
behavior, and is so little used that no one has been interested in fixing it.
To ensure that possible uses are covered, remove the ALIAS declaration's
arbitrary restriction that only $n identifiers can be aliased.
(We could alternatively make RENAME act just like ALIAS, but per discussion
having two different ways to do the same thing is probably more confusing than
helpful.)
Tom Lane [Thu, 5 Nov 2009 04:38:29 +0000 (04:38 +0000)]
Allow binary-coercible cases in ri_HashCompareOp; there are some such cases
that are not handled by find_coercion_pathway, notably composite->RECORD.
Now that 8.4 supports composites as primary keys, it's worth dealing with
this case.
Tom Lane [Wed, 4 Nov 2009 23:47:04 +0000 (23:47 +0000)]
Rename some encoding conversion modules to keep pathnames in our source
tarballs under 100 characters. This should avoid failures with certain
untarring tools (WinZip and Midnight Commander have been mentioned as
likely suspects). Per my proposal of yesterday.
catversion bumped since the initial contents of pg_proc change.
Tom Lane [Wed, 4 Nov 2009 23:15:08 +0000 (23:15 +0000)]
Make expression locations for LIKE and SIMILAR TO constructs uniformly point
at the first keyword of the expression, rather than drawing a rather
artificial distinction between the ESCAPE subclause and the rest.
Per gripe from Gokulakannan Somasundaram and subsequent discusssion.
Tom Lane [Wed, 4 Nov 2009 22:26:08 +0000 (22:26 +0000)]
Add support for invoking parser callback hooks via SPI and in cached plans.
As proof of concept, modify plpgsql to use the hooks. plpgsql is still
inserting $n symbols textually, but the "back end" of the parsing process now
goes through the ParamRef hook instead of using a fixed parameter-type array,
and then execution only fetches actually-referenced parameters, using a hook
added to ParamListInfo.
Although there's a lot left to be done in plpgsql, this already cures the
"if (TG_OP = 'INSERT' and NEW.foo ...)" problem, as illustrated by the
changed regression test.
Heikki Linnakangas [Wed, 4 Nov 2009 12:51:30 +0000 (12:51 +0000)]
Disable triggering failover with a signal in pg_standby on Windows, because
Windows doesn't do signal processing like other platforms do. It never
really worked, but recent changes to the signal handling made it crash.
This fixes bug #4961. Patch by Fujii Masao.
Heikki Linnakangas [Wed, 4 Nov 2009 12:24:23 +0000 (12:24 +0000)]
Allow rewriting ALTER TABLE to skip WAL logging.
Itagaki Takahiro, with small changes by me and Simon.
Peter Eisentraut [Tue, 3 Nov 2009 21:28:10 +0000 (21:28 +0000)]
Build bzip2 tarball in dist target as well
Peter Eisentraut [Tue, 3 Nov 2009 14:52:10 +0000 (14:52 +0000)]
Fix regression tests for psql \d view patch
Peter Eisentraut [Tue, 3 Nov 2009 11:05:03 +0000 (11:05 +0000)]
Improve PL/Python elog output
When the elog functions (plpy.info etc.) get a single argument, just print
that argument instead of printing the single-member tuple like ('foo',).
Peter Eisentraut [Tue, 3 Nov 2009 10:34:47 +0000 (10:34 +0000)]
In psql, show view definition only with \d+, not with \d
The rationale is that view definitions tend to be long and obscure the
main information about the view.
Peter Eisentraut [Tue, 3 Nov 2009 09:35:18 +0000 (09:35 +0000)]
Fix obscure segfault condition in PL/Python
In PLy_output(), when the elog() call in the TRY branch throws an exception
(this can happen when a statement timeout kicks in, for example), the
PyErr_SetString() call in the CATCH branch can cause a segfault, because the
Py_XDECREF(so) call before it releases memory that is still used by the sv
variable that PyErr_SetString() uses as argument, because sv points into
memory owned by so.
Backpatched back to 8.0, where this code was introduced.
I also threw in a couple of volatile declarations for variables that are used
before and after the TRY. I don't think they caused the crash that I
observed, but they could become issues.
Tom Lane [Sun, 1 Nov 2009 22:30:54 +0000 (22:30 +0000)]
Dept of second thoughts: after studying index_getnext() a bit more I realize
that it can scribble on scan->xs_ctup.t_self while following HOT chains,
so we can't rely on that to stay valid between hashgettuple() calls.
Introduce a private variable in HashScanOpaque, instead.
Tom Lane [Sun, 1 Nov 2009 21:25:25 +0000 (21:25 +0000)]
Fix two serious bugs introduced into hash indexes by the 8.4 patch that made
hash indexes keep entries sorted by hash value. First, the original plans for
concurrency assumed that insertions would happen only at the end of a page,
which is no longer true; this could cause scans to transiently fail to find
index entries in the presence of concurrent insertions. We can compensate
by teaching scans to re-find their position after re-acquiring read locks.
Second, neither the bucket split nor the bucket compaction logic had been
fixed to preserve hashvalue ordering, so application of either of those
processes could lead to permanent corruption of an index, in the sense
that searches might fail to find entries that are present.
This patch fixes the split and compaction logic to preserve hashvalue
ordering, but it cannot do anything about pre-existing corruption. We will
need to recommend reindexing all hash indexes in the 8.4.2 release notes.
To buy back the performance loss hereby induced in split and compaction,
fix them to use PageIndexMultiDelete instead of retail PageIndexDelete
operations. We might later want to do something with qsort'ing the
page contents rather than doing a binary search for each insertion,
but that seemed more invasive than I cared to risk in a back-patch.
Per bug #5157 from Jeff Janes and subsequent investigation.
Tom Lane [Sat, 31 Oct 2009 18:11:59 +0000 (18:11 +0000)]
Ensure the previous Perl interpreter selection is restored upon exit from
plperl_call_handler, in both the normal and error-exit paths. Per report
from Alexey Klyukin.
Tom Lane [Sat, 31 Oct 2009 01:41:31 +0000 (01:41 +0000)]
Implement parser hooks for processing ColumnRef and ParamRef nodes, as per my
recent proposal. As proof of concept, remove knowledge of Params from the
core parser, arranging for them to be handled entirely by parser hook
functions. It turns out we need an additional hook for that --- I had
forgotten about the code that handles inferring a parameter's type from
context.
This is a preliminary step towards letting plpgsql handle its variables
through parser hooks. Additional work remains to be done to expose the
facility through SPI, but I think this is all the changes needed in the core
parser.
Tom Lane [Fri, 30 Oct 2009 20:58:45 +0000 (20:58 +0000)]
Make the overflow guards in ExecChooseHashTableSize be more protective.
The original coding ensured nbuckets and nbatch didn't exceed INT_MAX,
which while not insane on its own terms did nothing to protect subsequent
code like "palloc(nbatch * sizeof(BufFile *))". Since enormous join size
estimates might well be planner error rather than reality, it seems best
to constrain the initial sizes to be not more than work_mem/sizeof(pointer),
thus ensuring the allocated arrays don't exceed work_mem. We will allow
nbatch to get bigger than that during subsequent ExecHashIncreaseNumBatches
calls, but we should still guard against integer overflow in those palloc
requests. Per bug #5145 from Bernt Marius Johnsen.
Although the given test case only seems to fail back to 8.2, previous
releases have variants of this issue, so patch all supported branches.
Peter Eisentraut [Thu, 29 Oct 2009 21:57:17 +0000 (21:57 +0000)]
Remove some leftovers of split tarball support
Tom Lane [Wed, 28 Oct 2009 18:51:56 +0000 (18:51 +0000)]
Un-break EXPLAIN for Append plans. I messed this up a few days ago while
adding the ModifyTable node type --- I had been thinking ModifyTable should
replace Append as a special case in push_plan(), but actually both of them
have to be special-cased.
Tom Lane [Wed, 28 Oct 2009 18:09:44 +0000 (18:09 +0000)]
Fix \df to re-allow regexp special characters in the function name pattern.
This has always worked, up until somebody's thinko here:
http://archives.postgresql.org/pgsql-committers/2009-04/msg00233.php
Per bug #5143 from Piotr Wolinski.
Tom Lane [Wed, 28 Oct 2009 17:36:50 +0000 (17:36 +0000)]
Fix AcquireRewriteLocks to be sure that it acquires the right lock strength
when FOR UPDATE is propagated down into a sub-select expanded from a view.
Similar bug to parser's isLockedRel issue that I fixed yesterday; likewise
seems not quite worth the effort to back-patch.
Tom Lane [Wed, 28 Oct 2009 14:55:47 +0000 (14:55 +0000)]
When FOR UPDATE/SHARE is used with LIMIT, put the LockRows plan node
underneath the Limit node, not atop it. This fixes the old problem that such
a query might unexpectedly return fewer rows than the LIMIT says, due to
LockRows discarding updated rows.
There is a related problem that LockRows might destroy the sort ordering
produced by earlier steps; but fixing that by pushing LockRows below Sort
would create serious performance problems that are unjustified in many
real-world applications, as well as potential deadlock problems from locking
many more rows than expected. Instead, keep the present semantics of applying
FOR UPDATE after ORDER BY within a single query level; but allow the user to
specify the other way by writing FOR UPDATE in a sub-select. To make that
work, track whether FOR UPDATE appeared explicitly in sub-selects or got
pushed down from the parent, and don't flatten a sub-select that contained an
explicit FOR UPDATE.
Tom Lane [Tue, 27 Oct 2009 20:14:27 +0000 (20:14 +0000)]
Fix AfterTriggerSaveEvent to use a test and elog, not just Assert, to check
that it's called within an AfterTriggerBeginQuery/AfterTriggerEndQuery pair.
The RI cascade triggers suppress that overhead on the assumption that they
are always run non-deferred, so it's possible to violate the condition if
someone mistakenly changes pg_trigger to mark such a trigger deferred.
We don't really care about supporting that, but throwing an error instead
of crashing seems desirable. Per report from Marcelo Costa.
Tom Lane [Tue, 27 Oct 2009 17:11:18 +0000 (17:11 +0000)]
Make FOR UPDATE/SHARE in the primary query not propagate into WITH queries;
for example in
WITH w AS (SELECT * FROM foo) SELECT * FROM w, bar ... FOR UPDATE
the FOR UPDATE will now affect bar but not foo. This is more useful and
consistent than the original 8.4 behavior, which tried to propagate FOR UPDATE
into the WITH query but always failed due to assorted implementation
restrictions. Even though we are in process of removing those restrictions,
it seems correct on philosophical grounds to not let the outer query's
FOR UPDATE affect the WITH query.
In passing, fix isLockedRel which frequently got things wrong in
nested-subquery cases: "FOR UPDATE OF foo" applies to an alias foo in the
current query level, not subqueries. This has been broken for a long time,
but it doesn't seem worth back-patching further than 8.4 because the actual
consequences are minimal. At worst the parser would sometimes get
RowShareLock on a relation when it should be AccessShareLock or vice versa.
That would only make a difference if someone were using ExclusiveLock
concurrently, which no standard operation does, and anyway FOR UPDATE
doesn't result in visible changes so it's not clear that the someone would
notice any problem. Between that and the fact that FOR UPDATE barely works
with subqueries at all in existing releases, I'm not excited about worrying
about it.
Alvaro Herrera [Tue, 27 Oct 2009 13:58:28 +0000 (13:58 +0000)]
Fix documentation on the toast.fillfactor reloption: it doesn't exist.
Per note from Zoltan Boszormenyi.
Peter Eisentraut [Mon, 26 Oct 2009 21:33:01 +0000 (21:33 +0000)]
Simplify a few makefile rules since install-sh can now install multiple
files in one run.
Peter Eisentraut [Mon, 26 Oct 2009 21:11:22 +0000 (21:11 +0000)]
Check errors in for loop
Heikki Linnakangas [Mon, 26 Oct 2009 16:13:11 +0000 (16:13 +0000)]
Fix range check in date_recv that tried to limit accepted values to only
those accepted by date_in(). I confused julian day numbers and number of
days since the postgres epoch 2000-01-01 in the original patch.
I just noticed that it's still easy to get such out-of-range values into
the database using to_date or +- operators, but this patch doesn't do
anything about those functions.
Per report from James Pye.
Tom Lane [Mon, 26 Oct 2009 02:26:45 +0000 (02:26 +0000)]
Re-implement EvalPlanQual processing to improve its performance and eliminate
a lot of strange behaviors that occurred in join cases. We now identify the
"current" row for every joined relation in UPDATE, DELETE, and SELECT FOR
UPDATE/SHARE queries. If an EvalPlanQual recheck is necessary, we jam the
appropriate row into each scan node in the rechecking plan, forcing it to emit
only that one row. The former behavior could rescan the whole of each joined
relation for each recheck, which was terrible for performance, and what's much
worse could result in duplicated output tuples.
Also, the original implementation of EvalPlanQual could not re-use the recheck
execution tree --- it had to go through a full executor init and shutdown for
every row to be tested. To avoid this overhead, I've associated a special
runtime Param with each LockRows or ModifyTable plan node, and arranged to
make every scan node below such a node depend on that Param. Thus, by
signaling a change in that Param, the EPQ machinery can just rescan the
already-built test plan.
This patch also adds a prohibition on set-returning functions in the
targetlist of SELECT FOR UPDATE/SHARE. This is needed to avoid the
duplicate-output-tuple problem. It seems fairly reasonable since the
other restrictions on SELECT FOR UPDATE are meant to ensure that there
is a unique correspondence between source tuples and result tuples,
which an output SRF destroys as much as anything else does.
Peter Eisentraut [Fri, 23 Oct 2009 05:24:52 +0000 (05:24 +0000)]
When querying a table with child tables, do not check permissions on the
child tables. This was found to be useless and confusing in virtually all
cases, and also contrary to the SQL standard.
Tom Lane [Wed, 21 Oct 2009 20:38:58 +0000 (20:38 +0000)]
Remove regex_flavor GUC, so that regular expressions are always "advanced"
style by default. Per discussion, there seems to be hardly anything that
really relies on being able to change the regex flavor, so the ability to
select it via embedded options ought to be enough for any stragglers.
Also, if we didn't remove the GUC, we'd really be morally obligated to
mark the regex functions non-immutable, which'd possibly create performance
issues.
Tom Lane [Wed, 21 Oct 2009 20:22:38 +0000 (20:22 +0000)]
Remove add_missing_from GUC and associated parser support for "implicit RTEs".
Per recent discussion, add_missing_from has been deprecated for long enough to
consider removing, and it's getting in the way of planned parser refactoring.
The system now always behaves as though add_missing_from were OFF.
Peter Eisentraut [Wed, 21 Oct 2009 19:43:06 +0000 (19:43 +0000)]
Finalize 8.5alpha2 release notes, with updates from Josh Berkus
Peter Eisentraut [Tue, 20 Oct 2009 19:52:58 +0000 (19:52 +0000)]
Preliminary release notes for 8.5alpha2
Peter Eisentraut [Tue, 20 Oct 2009 18:23:27 +0000 (18:23 +0000)]
Translations update for 8.5alpha2
Peter Eisentraut [Tue, 20 Oct 2009 18:22:19 +0000 (18:22 +0000)]
Update translation updating procedure. This pertains to some changes I
made to automatically exclude translations below the 80% minimum.
Magnus Hagander [Sat, 17 Oct 2009 05:14:52 +0000 (05:14 +0000)]
Fix typo in previous release as reported by Itagaki Takahiro, but missed
by me.
Magnus Hagander [Sat, 17 Oct 2009 00:24:51 +0000 (00:24 +0000)]
Write to the Windows eventlog in UTF16, converting the message encoding
as necessary.
Itagaki Takahiro with some changes from me
Tom Lane [Fri, 16 Oct 2009 22:08:36 +0000 (22:08 +0000)]
Rewrite pam_passwd_conv_proc to be more robust: avoid assuming that the
pam_message array contains exactly one PAM_PROMPT_ECHO_OFF message.
Instead, deal with however many messages there are, and don't throw error
for PAM_ERROR_MSG and PAM_TEXT_INFO messages. This logic is borrowed from
openssh 5.2p1, which hopefully has seen more real-world PAM usage than we
have. Per bug #5121 from Ryan Douglas, which turned out to be caused by
the conv_proc being called with zero messages. Apparently that is normal
behavior given the combination of Linux pam_krb5 with MS Active Directory
as the domain controller.
Patch all the way back, since this code has been essentially untouched
since 7.4. (Surprising we've not heard complaints before.)
Heikki Linnakangas [Fri, 16 Oct 2009 10:38:25 +0000 (10:38 +0000)]
FREEZE and VERBOSE options were in wrong order in the VACUUM command that
vacuumdb produces. Per report by Thom Brown.
Tom Lane [Thu, 15 Oct 2009 23:39:13 +0000 (23:39 +0000)]
First committed version of plpython_unicode_0.out did not actually contain the
required \200 bytes. Let's see if this commit works, or if CVS is messing it up.