Tom Lane [Wed, 29 May 2002 17:36:40 +0000 (17:36 +0000)]
Add a note about the interpretation of amcanmulticol and amindexnulls:
a multicolumn-capable index AM *must* support nulls in index columns
after the first one.
Tom Lane [Tue, 28 May 2002 23:56:51 +0000 (23:56 +0000)]
Rearrange LOG_CONNECTIONS code so that two log messages are made:
one immediately upon forking to handle a new connection, and one after
the authentication cycle is finished. Per today's pggeneral discussion.
Tom Lane [Tue, 28 May 2002 22:26:57 +0000 (22:26 +0000)]
Rework pg_dump namespace search criteria so that dumping of user objects
having names conflicting with system objects will work --- the search
path is now user-schema, pg_catalog rather than implicitly the other way
around. Note this requires being careful to explicitly qualify references
to system names whenever pg_catalog is not first in the search path.
Also, add support for dumping ACLs of schemas.
Tom Lane [Mon, 27 May 2002 19:53:33 +0000 (19:53 +0000)]
Distinguish between MaxHeapAttributeNumber and MaxTupleAttributeNumber,
where the latter is made slightly larger to allow for in-memory tuples
containing resjunk attributes. Responds to today's complaint that one
cannot UPDATE a table containing the allegedly-legal maximum number of
columns.
Also, apply Manfred Koizar's recent patch to avoid extra alignment padding
when there is a null bitmap. This saves bytes in some cases while not
creating any backward-compatibility problem AFAICS.
Tom Lane [Sat, 25 May 2002 20:00:12 +0000 (20:00 +0000)]
Remove AMI_OVERRIDE tests from tqual.c routines; they aren't necessary
and just slow down normal operations (only fractionally, but a cycle saved
is a cycle earned). Improve documentation of AMI_OVERRIDE behavior.
Tom Lane [Fri, 24 May 2002 18:57:57 +0000 (18:57 +0000)]
Mark index entries "killed" when they are no longer visible to any
transaction, so as to avoid returning them out of the index AM. Saves
repeated heap_fetch operations on frequently-updated rows. Also detect
queries on unique keys (equality to all columns of a unique index), and
don't bother continuing scan once we have found first match.
Killing is implemented in the btree and hash AMs, but not yet in rtree
or gist, because there isn't an equally convenient place to do it in
those AMs (the outer amgetnext routine can't do it without re-pinning
the index page).
Did some small cleanup on APIs of HeapTupleSatisfies, heap_fetch, and
index_insert to make this a little easier.
Peter Eisentraut [Fri, 24 May 2002 18:10:17 +0000 (18:10 +0000)]
Change PL/Tcl build to use configured compiler and Makefile.shlib
system, not Tcl-provided one.
Make sure export file, if any, is cleaned.
Tcl configuration is now read directly in configure and recorded in
Makefile.global. This eliminates some duplicate efforts and allows
for easier hand-editing of the results, if necessary.
Tom Lane [Wed, 22 May 2002 21:40:55 +0000 (21:40 +0000)]
Modify sequence state storage to eliminate dangling-pointer problem
exemplified by bug #671. Moving the storage to relcache turned out to
be a bad idea because relcache might decide to discard the info. Instead,
open and close the relcache entry on each sequence operation, and use
a record of the current XID to discover whether we already hold
AccessShareLock on the sequence.
Peter Eisentraut [Wed, 22 May 2002 17:21:02 +0000 (17:21 +0000)]
Add optional "validator" function to languages that can validate the
function body (and other properties) as a function in the language
is created. This generalizes ad hoc code that already existed for
the built-in languages.
The validation now happens after the pg_proc tuple of the new function
is created, so it is possible to define recursive SQL functions.
Add some regression test cases that cover bogus function definition
attempts.
Tom Lane [Wed, 22 May 2002 15:57:40 +0000 (15:57 +0000)]
Make RelationForgetRelation error out if the relcache entry has nonzero
reference count. This avoids leaving dangling pointers around, as in
recent bug report against sequences (bug# 671).
Hiroshi Inoue [Wed, 22 May 2002 05:51:03 +0000 (05:51 +0000)]
1) Support Keyset Driven driver cursors.
2) Supprt ARD precision/scale and SQL_C_NUEMRIC.
3) Minimal implementation of SQLGetDiagField().
4) SQLRowCount() reports the result of SQLSetPos and SQLBulkOperation.
5) int8 -> SQL_NUMERIC for Microsoft Jet.
6) Support isolation level change.
7) ODBC3.0 SQLSTATE code.
8) Append mode log files.
Tom Lane [Tue, 21 May 2002 22:59:01 +0000 (22:59 +0000)]
Since COPY fires triggers, it seems like a good idea for it to use
a frozen (copied) snapshot too. Move execMain's snapshot copying code
out into a subroutine in case we find other places that need it.
Tom Lane [Tue, 21 May 2002 22:18:08 +0000 (22:18 +0000)]
Remove SetQuerySnapshot() from FETCH processing. No longer necessary
or appropriate, since the snapshot that will be used by the cursor was
frozen when ExecutorStart was run for it.
Tom Lane [Tue, 21 May 2002 22:05:55 +0000 (22:05 +0000)]
Remove global variable scanCommandId in favor of storing a command ID
in snapshots, per my proposal of a few days ago. Also, tweak heapam.c
routines (heap_insert, heap_update, heap_delete, heap_mark4update) to
be passed the command ID to use, instead of doing GetCurrentCommandID.
For catalog updates they'll still get passed current command ID, but
for updates generated from the main executor they'll get passed the
command ID saved in the snapshot the query is using. This should fix
some corner cases associated with functions and triggers that advance
current command ID while an outer query is still in progress.
Tom Lane [Mon, 20 May 2002 23:51:44 +0000 (23:51 +0000)]
Restructure indexscan API (index_beginscan, index_getnext) per
yesterday's proposal to pghackers. Also remove unnecessary parameters
to heap_beginscan, heap_rescan. I modified pg_proc.h to reflect the
new numbers of parameters for the AM interface routines, but did not
force an initdb because nothing actually looks at those fields.
Tom Lane [Sat, 18 May 2002 18:49:41 +0000 (18:49 +0000)]
Teach query_tree_walker, query_tree_mutator, and SS_finalize_plan to
process function RTE expressions, which they were previously missing.
This allows outer-Var references and subselects to work correctly in
the arguments of a function RTE. Install check to prevent function RTEs
from cross-referencing Vars of sibling FROM-items, which doesn't make
any sense (if you want to join, write a JOIN or WHERE clause).
Tom Lane [Sat, 18 May 2002 02:25:50 +0000 (02:25 +0000)]
Change set_plan_references and join_references to take an rtable List
rather than a Query node; this allows set_plan_references to recurse
into subplans correctly. Fixes core dump on full outer joins in
subplans. Also, invoke preprocess_expression on function RTEs'
function expressions. This seems to fix the planner's problems with
outer-level Vars in function RTEs.
Tom Lane [Fri, 17 May 2002 20:53:33 +0000 (20:53 +0000)]
Support temporary setting of search path during CREATE SCHEMA; this
allows the example in the CREATE SCHEMA ref page to actually work now.
Also, clean up when the transaction that initially creates a temp-table
namespace is later aborted. Simplify internal representation of search
path by folding special cases into the main list.
Tom Lane [Fri, 17 May 2002 20:32:29 +0000 (20:32 +0000)]
Dept. of second thoughts: interaction between DoIt and makeDepend
in set_config_option wasn't quite right. Also clean up a couple other
things that could have been done better.
Tom Lane [Fri, 17 May 2002 01:19:19 +0000 (01:19 +0000)]
Merge the last few variable.c configuration variables into the generic
GUC support. It's now possible to set datestyle, timezone, and
client_encoding from postgresql.conf and per-database or per-user
settings. Also, implement rollback of SET commands that occur in a
transaction that later fails. Create a SET LOCAL var = value syntax
that sets the variable only for the duration of the current transaction.
All per previous discussions in pghackers.
Tom Lane [Tue, 14 May 2002 18:16:51 +0000 (18:16 +0000)]
Remove unnecessary pfree's in geometric operators. At least one of these
is actively dangerous, per bug report from Ewald Geschwinde 14-May-02,
and several of the rest look suspicious to me. Since there is no longer
any significant value in retail pfree's in these functions, just get
rid of all of them for safety's sake.
Peter Eisentraut [Tue, 14 May 2002 13:05:43 +0000 (13:05 +0000)]
Make regression tests locale-proof by setting some locale categories
to C at run-time, and providing alternative output files for different
sort orders.
Barry Lind [Tue, 14 May 2002 03:00:35 +0000 (03:00 +0000)]
fixed problem connecting to server with client_min_messages set to debug. The code was not expecting to receive notice messages during the connection handshake.
Tom Lane [Mon, 13 May 2002 17:45:30 +0000 (17:45 +0000)]
Make operators have their own comments separate from those of the
underlying function; but cause psql's \do to show the underlying
function's comment if the operator has no comment of its own, to preserve
the useful functionality of the original behavior. Also, implement
COMMENT ON SCHEMA. Patch from Rod Taylor.
Tom Lane [Sun, 12 May 2002 23:43:04 +0000 (23:43 +0000)]
Get rid of long-since-vestigial Iter node type, in favor of adding a
returns-set boolean field in Func and Oper nodes. This allows cleaner,
more reliable tests for expressions returning sets in the planner and
parser. For example, a WHERE clause returning a set is now detected
and complained of in the parser, not only at runtime.
Tom Lane [Sun, 12 May 2002 20:10:05 +0000 (20:10 +0000)]
First pass at set-returning-functions in FROM, by Joe Conway with
some kibitzing from Tom Lane. Not everything works yet, and there's
no documentation or regression test, but let's commit this so Joe
doesn't need to cope with tracking changes in so many files ...
Tom Lane [Mon, 6 May 2002 19:47:30 +0000 (19:47 +0000)]
Accept SET SESSION AUTHORIZATION DEFAULT and RESET SESSION AUTHORIZATION
to reset session userid to the originally-authenticated name. Also,
relax SET SESSION AUTHORIZATION to allow specifying one's own username
even if one is not superuser, so as to avoid unnecessary error messages
when loading a pg_dump file that uses this command. Per discussion from
several months ago.
Tom Lane [Mon, 6 May 2002 02:39:01 +0000 (02:39 +0000)]
Reorder snapshot checks to save a couple comparisons in the common case,
where the tuple's xmin or xmax is older than the snapshot xmin. There
is no need to check it against snapshot xmax in that case.
Tom Lane [Sun, 5 May 2002 00:03:29 +0000 (00:03 +0000)]
Create an internal semaphore API that is not tied to SysV semaphores.
As proof of concept, provide an alternate implementation based on POSIX
semaphores. Also push the SysV shared-memory implementation into a
separate file so that it can be replaced conveniently.
Tom Lane [Fri, 3 May 2002 20:15:02 +0000 (20:15 +0000)]
Make ruleutils.c schema-aware. Displayed names are schema-qualified
only if they would not be found without qualification given the current
search path, as per idea from Peter Eisentraut.
Tom Lane [Fri, 3 May 2002 00:32:19 +0000 (00:32 +0000)]
Retire xlateSqlType/xlateSqlFunc; all type name translations are now
handled as special productions. This is needed to keep us honest about
user-schema type names that happen to coincide with system type names.
Per pghackers discussion 24-Apr. To avoid bloating the keyword list
too much, I removed the translations for datetime, timespan, and lztext,
all of which were slated for destruction several versions back anyway.
Tom Lane [Thu, 2 May 2002 18:44:11 +0000 (18:44 +0000)]
Tweak scanner/grammar interface so that the keyword-as-identifier rules
in gram.y can make use of the keywords.c string table, instead of having
their own copies of the keyword strings. This saves a few kilobytes and
more importantly eliminates an opportunity for cut-and-paste errors.
Tom Lane [Wed, 1 May 2002 23:06:41 +0000 (23:06 +0000)]
Add routines in namespace.c to determine whether objects are visible
in the search path. (We might want to make these available as SQL
functions too, but I haven't done that yet.) Fix format_type to be
schema-aware.
Tom Lane [Wed, 1 May 2002 17:12:08 +0000 (17:12 +0000)]
Improve lexer's error reporting. You get the whole token mentioned now
in parse error messages, not just the part scanned by the last flex rule.
For example,
select "foo" "bar";
used to draw
ERROR: parser: parse error at or near """
which was rather unhelpful. Now it gives
ERROR: parser: parse error at or near ""bar""
Also, error messages concerning bitstring literals and suchlike will
quote the source text at you, not the processed internal form of the literal.