]> granicus.if.org Git - postgresql/log
postgresql
21 years agoChange CREATE TABLE AS / SELECT INTO to create the new table with OIDs,
Tom Lane [Thu, 23 Jan 2003 05:10:41 +0000 (05:10 +0000)]
Change CREATE TABLE AS / SELECT INTO to create the new table with OIDs,
for backwards compatibility with pre-7.3 behavior.  Per discussion on
pgsql-general and pgsql-hackers.

21 years agoUpdate 'Mathematical Functions' table to reflect 7.3 addition of float8
Tom Lane [Thu, 23 Jan 2003 01:22:59 +0000 (01:22 +0000)]
Update 'Mathematical Functions' table to reflect 7.3 addition of float8
forms of ceil(), floor(), sign().  Back-patch this and other recent
doc fixes in this file to the 7.3 branch.

21 years agoFix sloppy comment.
Tom Lane [Wed, 22 Jan 2003 20:44:20 +0000 (20:44 +0000)]
Fix sloppy comment.

21 years agoMake estimation of mergejoin scan selectivities more robust, per recent
Tom Lane [Wed, 22 Jan 2003 20:16:42 +0000 (20:16 +0000)]
Make estimation of mergejoin scan selectivities more robust, per recent
example from RaÇl GutiÅrrez.

21 years agoImprove print_expr() a little. It's still not very bright though.
Tom Lane [Wed, 22 Jan 2003 19:26:35 +0000 (19:26 +0000)]
Improve print_expr() a little.  It's still not very bright though.

21 years agoImplement choice between hash-based and sort-based grouping for doing
Tom Lane [Wed, 22 Jan 2003 00:07:00 +0000 (00:07 +0000)]
Implement choice between hash-based and sort-based grouping for doing
DISTINCT processing on the output of an IN sub-select.

21 years agoFix coredump problem in plpgsql's RETURN NEXT. When a SELECT INTO
Tom Lane [Tue, 21 Jan 2003 22:06:12 +0000 (22:06 +0000)]
Fix coredump problem in plpgsql's RETURN NEXT.  When a SELECT INTO
that's selecting into a RECORD variable returns zero rows, make it
assign an all-nulls row to the RECORD; this is consistent with what
happens when the SELECT INTO target is not a RECORD.  In support of
this, tweak the SPI code so that a valid tuple descriptor is returned
even when a SPI select returns no rows.

21 years agoSynced parser.
Michael Meskes [Tue, 21 Jan 2003 20:01:12 +0000 (20:01 +0000)]
Synced parser.

21 years agoAvoid gratuitous variation in spelling of same error message.
Tom Lane [Tue, 21 Jan 2003 19:44:26 +0000 (19:44 +0000)]
Avoid gratuitous variation in spelling of same error message.

21 years agoFix sed expression for BSD sed.
Peter Eisentraut [Tue, 21 Jan 2003 10:11:52 +0000 (10:11 +0000)]
Fix sed expression for BSD sed.

21 years agoIN clauses appearing at top level of WHERE can now be handled as joins.
Tom Lane [Mon, 20 Jan 2003 18:55:07 +0000 (18:55 +0000)]
IN clauses appearing at top level of WHERE can now be handled as joins.
There are two implementation techniques: the executor understands a new
JOIN_IN jointype, which emits at most one matching row per left-hand row,
or the result of the IN's sub-select can be fed through a DISTINCT filter
and then joined as an ordinary relation.
Along the way, some minor code cleanup in the optimizer; notably, break
out most of the jointree-rearrangement preprocessing in planner.c and
put it in a new file prep/prepjointree.c.

21 years agoThis patch includes a lot of minor cleanups to the SGML documentation,
Bruce Momjian [Sun, 19 Jan 2003 00:13:31 +0000 (00:13 +0000)]
This patch includes a lot of minor cleanups to the SGML documentation,
including:

- replacing all the appropriate usages of <citetitle>PostgreSQL
...</citetitle> with &cite-user;, &cite-admin;, and so on

- fix an omission in the EXECUTE documentation

- add some more text to the EXPLAIN documentation

- improve the PL/PgSQL RETURN NEXT documentation (more work to do here)

- minor markup fixes

Neil Conway

21 years agoImprove symlink handling for C tags file.
Bruce Momjian [Sat, 18 Jan 2003 06:06:51 +0000 (06:06 +0000)]
Improve symlink handling for C tags file.

21 years agoAdd section on testing index scan.
Bruce Momjian [Fri, 17 Jan 2003 17:15:14 +0000 (17:15 +0000)]
Add section on testing index scan.

21 years agoAdd dash to:
Bruce Momjian [Fri, 17 Jan 2003 03:28:18 +0000 (03:28 +0000)]
Add dash to:

     set-returning functions or SRF's.

for clarity.

21 years agoFix parse_agg.c to detect ungrouped Vars in sub-SELECTs; remove code
Tom Lane [Fri, 17 Jan 2003 03:25:04 +0000 (03:25 +0000)]
Fix parse_agg.c to detect ungrouped Vars in sub-SELECTs; remove code
that used to do it in planner.  That was an ancient kluge that was
never satisfactory; errors should be detected at parse time when possible.
But at the time we didn't have the support mechanism (expression_tree_walker
et al) to make it convenient to do in the parser.

21 years agoAdjust API of expression_tree_mutator and query_tree_mutator to
Tom Lane [Fri, 17 Jan 2003 02:01:21 +0000 (02:01 +0000)]
Adjust API of expression_tree_mutator and query_tree_mutator to
simplify callers.  It turns out the common case is that the caller
does want to recurse into sub-queries, so push support for that into
these subroutines.

21 years agoAdd code to print information about a detected deadlock cycle. The
Tom Lane [Thu, 16 Jan 2003 21:01:45 +0000 (21:01 +0000)]
Add code to print information about a detected deadlock cycle.  The
printed data is comparable to what you could read in the pg_locks view,
were you fortunate enough to have been looking at it at the right time.

21 years agoOoops, didn't cut-and-paste quite enough code from ResolveNew;
Tom Lane [Thu, 16 Jan 2003 18:26:02 +0000 (18:26 +0000)]
Ooops, didn't cut-and-paste quite enough code from ResolveNew;
with result that flatten_join_alias_vars failed to descend into subselects.

21 years agoFix bogus formatting of DROP DATABASE command, per Christopher Kings-Lynne.
Tom Lane [Thu, 16 Jan 2003 15:27:59 +0000 (15:27 +0000)]
Fix bogus formatting of DROP DATABASE command, per Christopher Kings-Lynne.

21 years agoRepair an embarrassingly large number of alphabetization mistakes in the
Tom Lane [Thu, 16 Jan 2003 00:26:49 +0000 (00:26 +0000)]
Repair an embarrassingly large number of alphabetization mistakes in the
datetime token tables.  Even more embarrassing, the regression tests
revealed some of the problems --- but evidently the bogus output wasn't
questioned.  Add code to postmaster startup to directly check the tables
for correct ordering, in hopes of not being embarrassed like this again.

21 years agoAdd remaining documentation tables to information schema.
Peter Eisentraut [Wed, 15 Jan 2003 23:37:28 +0000 (23:37 +0000)]
Add remaining documentation tables to information schema.

21 years agoNow that switch_outer processing no longer relies on being run after
Tom Lane [Wed, 15 Jan 2003 23:10:32 +0000 (23:10 +0000)]
Now that switch_outer processing no longer relies on being run after
join_references(), it's practical to consolidate all join_references()
processing into the set_plan_references traversal in setrefs.c.  This
seems considerably cleaner than the old way where we did it for join
quals in createplan.c and for targetlists in setrefs.c.

21 years agoOops, put back changes. Those were Peter's, not mine.
Bruce Momjian [Wed, 15 Jan 2003 21:55:52 +0000 (21:55 +0000)]
Oops, put back changes.  Those were Peter's, not mine.

21 years agoRevert my changes to features.sgml.
Bruce Momjian [Wed, 15 Jan 2003 21:44:35 +0000 (21:44 +0000)]
Revert my changes to features.sgml.

21 years agoMore comment cleanups for shared_buffers in postgresql.conf.
Bruce Momjian [Wed, 15 Jan 2003 20:49:56 +0000 (20:49 +0000)]
More comment cleanups for shared_buffers in postgresql.conf.

21 years agoUpdate shared_buffer comment in postgresql.conf.
Bruce Momjian [Wed, 15 Jan 2003 20:41:18 +0000 (20:41 +0000)]
Update shared_buffer comment in postgresql.conf.

21 years agoBack out ORDER BY addition to Inet, for Tom.
Bruce Momjian [Wed, 15 Jan 2003 20:01:01 +0000 (20:01 +0000)]
Back out ORDER BY addition to Inet, for Tom.

21 years agoAllow merge and hash joins to occur on arbitrary expressions (anything not
Tom Lane [Wed, 15 Jan 2003 19:35:48 +0000 (19:35 +0000)]
Allow merge and hash joins to occur on arbitrary expressions (anything not
containing a volatile function), rather than only on 'Var = Var' clauses
as before.  This makes it practical to do flatten_join_alias_vars at the
start of planning, which in turn eliminates a bunch of klugery inside the
planner to deal with alias vars.  As a free side effect, we now detect
implied equality of non-Var expressions; for example in
SELECT ... WHERE a.x = b.y and b.y = 42
we will deduce a.x = 42 and use that as a restriction qual on a.  Also,
we can remove the restriction introduced 12/5/02 to prevent pullup of
subqueries whose targetlists contain sublinks.
Still TODO: make statistical estimation routines in selfuncs.c and costsize.c
smarter about expressions that are more complex than plain Vars.  The need
for this is considerably greater now that we have to be able to estimate
the suitability of merge and hash join techniques on such expressions.

21 years agoReorder VARCHAR() to appear before CHAR() in docs.
Bruce Momjian [Wed, 15 Jan 2003 18:01:05 +0000 (18:01 +0000)]
Reorder VARCHAR() to appear before CHAR() in docs.

21 years agoThis patch fixes an incorrect statement and makes a few cleanups to
Bruce Momjian [Wed, 15 Jan 2003 16:45:17 +0000 (16:45 +0000)]
This patch fixes an incorrect statement and makes a few cleanups to
contrib/fulltextindex/README.fti

Backpatched to 7.3.X too.

Neil Conway

21 years agoThis patch includes some minor fixes and improvements to the SGML docs
Bruce Momjian [Wed, 15 Jan 2003 16:40:24 +0000 (16:40 +0000)]
This patch includes some minor fixes and improvements to the SGML docs
for PL/PgSQL.

Neil Conway

21 years agoAdd ORDER BY for inet regression, pointed out by Rod Taylor.
Bruce Momjian [Wed, 15 Jan 2003 16:35:50 +0000 (16:35 +0000)]
Add ORDER BY for inet regression, pointed out by Rod Taylor.

21 years agoAdd sql_features table to information schema. Generate the features list
Peter Eisentraut [Tue, 14 Jan 2003 23:19:34 +0000 (23:19 +0000)]
Add sql_features table to information schema.  Generate the features list
in the documentation from that same data.

21 years agoReadd #include netinet/in.h for FreeBSD.
Bruce Momjian [Tue, 14 Jan 2003 22:52:57 +0000 (22:52 +0000)]
Readd #include netinet/in.h for FreeBSD.

21 years agoEscape ampersand.
Peter Eisentraut [Tue, 14 Jan 2003 10:19:02 +0000 (10:19 +0000)]
Escape ampersand.

21 years agoBumped up build number to 200
Barry Lind [Tue, 14 Jan 2003 09:42:07 +0000 (09:42 +0000)]
Bumped up build number to 200

21 years agoPatch from Florian Wunderlich to correctly support java Timestamps. Previously
Barry Lind [Tue, 14 Jan 2003 09:13:51 +0000 (09:13 +0000)]
Patch from Florian Wunderlich to correctly support java Timestamps.  Previously
the code would only capture milliseconds where as both postgres and the java
Timestamp object support greater resolution.
Also fixed a bug reported by Rhett Sutphin where the last digit of the
fractional seconds was lost when using timestamp without time zone

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1ResultSet.java
  jdbc/org/postgresql/test/jdbc2/TimestampTest.java

21 years agoPatch from Aaron Mulder to have pooled connections implement PGConnection
Barry Lind [Tue, 14 Jan 2003 05:46:49 +0000 (05:46 +0000)]
Patch from Aaron Mulder to have pooled connections implement PGConnection

 Modified Files:
  jdbc/org/postgresql/jdbc2/optional/PooledConnectionImpl.java
  jdbc/org/postgresql/test/jdbc2/optional/BaseDataSourceTest.java

21 years agoPatch from Kris Jurka to improve the performance of getImportedKeys().
Barry Lind [Tue, 14 Jan 2003 05:05:26 +0000 (05:05 +0000)]
Patch from Kris Jurka to improve the performance of getImportedKeys().
Use explicit joins to avoid using the genetic query optimizer.  Also fixed
a regression test that was failing to compile.  This change also cleans up
how key names are reported as per:
A change to the value of the FK_NAME column.  Currently the returned value
is the triggers arguments which look like

"<unnamed>\000t2\000t1\000UNSPECIFIED\000a\000a\000"

This was required for server versions < 7.3 when a user did not supply
constraint names.  Every constraint was named "<unnamed>"
.  7.3 has enforced unique constraint names per table so unnamed foreign
keys will have different names "$1", "$2" and so on.  I've used logic
along the lines of the following to preserve the unique names in the
original scheme, but allow people who go to the trouble of naming their
constraints to see them:

if (triggerargs.startsWith("<unnamed>")) {
fkname = [the whole ugly trigger args name originally used];
} else {
fkname = [the actual fk name];
}

 Modified Files:
  jdbc/org/postgresql/jdbc1/AbstractJdbc1DatabaseMetaData.java
  jdbc/org/postgresql/test/jdbc2/optional/ConnectionPoolTest.java

21 years agoAdd:
Bruce Momjian [Tue, 14 Jan 2003 00:47:35 +0000 (00:47 +0000)]
Add:

>  o Allow PL/PgSQL to support array element assignment

21 years agoAdd Hugarian FAQ, from Laszlo Hornyak
Bruce Momjian [Mon, 13 Jan 2003 23:00:32 +0000 (23:00 +0000)]
Add Hugarian FAQ, from Laszlo Hornyak

21 years agoFix information_schema.sql install to work when building outside
Tom Lane [Mon, 13 Jan 2003 20:37:18 +0000 (20:37 +0000)]
Fix information_schema.sql install to work when building outside
source tree.

21 years agoReconsider mechanism for marking sub-selects that are at top level of
Tom Lane [Mon, 13 Jan 2003 18:10:53 +0000 (18:10 +0000)]
Reconsider mechanism for marking sub-selects that are at top level of
a qualification clause (and hence can get away with being sloppy about
distinguishing FALSE from UNKNOWN).  We need to know this in subselect.c;
marking the subplans in setrefs.c is too late.

21 years agoChange Adjust_lo_type() so that it doesn't cause an error
Hiroshi Inoue [Mon, 13 Jan 2003 04:28:55 +0000 (04:28 +0000)]
Change Adjust_lo_type() so that it doesn't cause an error
even when cast functions are allowed to be volatile.

21 years agoFixes to pgcvslog for last narrive entry.
Bruce Momjian [Mon, 13 Jan 2003 01:57:47 +0000 (01:57 +0000)]
Fixes to pgcvslog for last narrive entry.

21 years agoCause planner to account for evaluation costs in targetlists and
Tom Lane [Mon, 13 Jan 2003 00:29:26 +0000 (00:29 +0000)]
Cause planner to account for evaluation costs in targetlists and
HAVING quals.  Normally this is an insignificant effect --- but it
will not be insignificant when these clauses contain sub-selects.
The added costs cannot affect the planning of the query containing
them, but they might have an impact when the query is a sub-query
of a larger one.

21 years agoRecent changes in sublink representation require exprType() to accept
Tom Lane [Mon, 13 Jan 2003 00:18:51 +0000 (00:18 +0000)]
Recent changes in sublink representation require exprType() to accept
SubPlan nodes, else explaining queries containing sublinks may fail.

21 years agoRevise cost_qual_eval() to compute both startup (one-time) and per-tuple
Tom Lane [Sun, 12 Jan 2003 22:35:29 +0000 (22:35 +0000)]
Revise cost_qual_eval() to compute both startup (one-time) and per-tuple
costs for expression evaluation, not only per-tuple cost as before.
This extension is needed in order to deal realistically with hashed or
materialized sub-selects.

21 years agoFix wrong/misleading comments, be more consistent about where to call
Tom Lane [Sun, 12 Jan 2003 22:01:38 +0000 (22:01 +0000)]
Fix wrong/misleading comments, be more consistent about where to call
ExecAssignResultTypeFromTL().

21 years agoFix some minor grammatical errors.
Tom Lane [Sun, 12 Jan 2003 18:42:59 +0000 (18:42 +0000)]
Fix some minor grammatical errors.

21 years agoUpdate CHAR().
Bruce Momjian [Sun, 12 Jan 2003 18:36:22 +0000 (18:36 +0000)]
Update CHAR().

21 years agoReplace RelidGetNamespaceId() by get_rel_namespace().
Peter Eisentraut [Sun, 12 Jan 2003 18:19:37 +0000 (18:19 +0000)]
Replace RelidGetNamespaceId() by get_rel_namespace().

21 years agoUpdate CHAR() description.
Bruce Momjian [Sun, 12 Jan 2003 14:58:46 +0000 (14:58 +0000)]
Update CHAR() description.

21 years agoUpdate date only.
Bruce Momjian [Sun, 12 Jan 2003 05:35:55 +0000 (05:35 +0000)]
Update date only.

21 years agoImprove CHAR() description.
Bruce Momjian [Sun, 12 Jan 2003 05:16:12 +0000 (05:16 +0000)]
Improve CHAR() description.

21 years agoFirst cut at implementing IN (and NOT IN) via hashtables. There is
Tom Lane [Sun, 12 Jan 2003 04:03:34 +0000 (04:03 +0000)]
First cut at implementing IN (and NOT IN) via hashtables.  There is
more to be done yet, but this is a good start.

21 years agoSGML build cleanups from Neil Conway.
Bruce Momjian [Sun, 12 Jan 2003 01:33:00 +0000 (01:33 +0000)]
SGML build cleanups from Neil Conway.

21 years agoFix capitalization.
Bruce Momjian [Sat, 11 Jan 2003 21:02:49 +0000 (21:02 +0000)]
Fix capitalization.

21 years agoBack out flockfile change for NetBSD. Giles Lean reports they are not
Bruce Momjian [Sat, 11 Jan 2003 19:38:23 +0000 (19:38 +0000)]
Back out flockfile change for NetBSD.  Giles Lean reports they are not
supported.

21 years agoCause symlinks for shared-library versioning to run in the standard
Tom Lane [Sat, 11 Jan 2003 17:22:19 +0000 (17:22 +0000)]
Cause symlinks for shared-library versioning to run in the standard
direction on HP-UX; our former approach has been obsolete since HPUX 9.

Giles Lean

21 years agoFix markup problem in link to other SGML file.
Bruce Momjian [Sat, 11 Jan 2003 17:03:45 +0000 (17:03 +0000)]
Fix markup problem in link to other SGML file.

21 years agoFix typo in an error message
Tatsuo Ishii [Sat, 11 Jan 2003 06:55:11 +0000 (06:55 +0000)]
Fix typo in an error message

21 years ago> > This patch improves the documentation for the shared_buffers GUC param.
Bruce Momjian [Sat, 11 Jan 2003 05:04:14 +0000 (05:04 +0000)]
> > This patch improves the documentation for the shared_buffers GUC param.
>
> I'd suggest that the runtime.sgml description explicitly say "values of
> at least a few thousand are recommended for production installations".

Neil Conway

21 years agoRewrite for-loop, because this is not the Obfuscated C Code Contest.
Bruce Momjian [Sat, 11 Jan 2003 05:01:03 +0000 (05:01 +0000)]
Rewrite for-loop, because this is not the Obfuscated C Code Contest.

Manfred Koizar

21 years agoA tiny patch to fix a typo in configure.in and another one in
Bruce Momjian [Sat, 11 Jan 2003 04:58:44 +0000 (04:58 +0000)]
A tiny patch to fix a typo in configure.in and another one in
RELEASE_CHANGES.

Manfred Koizar

21 years agoAdd doc links from SET to SET_CONSTRAINTS, SET_SESSION_AUTH,
Bruce Momjian [Sat, 11 Jan 2003 00:39:52 +0000 (00:39 +0000)]
Add doc links from SET to SET_CONSTRAINTS, SET_SESSION_AUTH,
SET_TRANSACTION.

21 years agoFix markup.
Peter Eisentraut [Sat, 11 Jan 2003 00:00:03 +0000 (00:00 +0000)]
Fix markup.

21 years agoCreate a new file executor/execGrouping.c to centralize utility routines
Tom Lane [Fri, 10 Jan 2003 23:54:24 +0000 (23:54 +0000)]
Create a new file executor/execGrouping.c to centralize utility routines
shared by nodeGroup, nodeAgg, and soon nodeSubplan.

21 years agoAdd missing <limits.h> for INT64_IS_BUSTED case.
Tom Lane [Fri, 10 Jan 2003 23:49:06 +0000 (23:49 +0000)]
Add missing <limits.h> for INT64_IS_BUSTED case.

21 years agoRead-only transactions, as defined in SQL.
Peter Eisentraut [Fri, 10 Jan 2003 22:03:30 +0000 (22:03 +0000)]
Read-only transactions, as defined in SQL.

21 years agoDo not shell-quote the name of the editor, so that the editor can be a
Peter Eisentraut [Fri, 10 Jan 2003 21:57:44 +0000 (21:57 +0000)]
Do not shell-quote the name of the editor, so that the editor can be a
command with arguments.

21 years agoFurther tweaking of parsetree & plantree representation of SubLinks.
Tom Lane [Fri, 10 Jan 2003 21:08:15 +0000 (21:08 +0000)]
Further tweaking of parsetree & plantree representation of SubLinks.
Simplify SubLink by storing just a List of operator OIDs, instead of
a list of incomplete OpExprs --- that was a bizarre and bulky choice,
with no redeeming social value since we have to build new OpExprs
anyway when forming the plan tree.

21 years agoAdd optional drop behavior clause to REVOKE command, for SQL conformance.
Peter Eisentraut [Fri, 10 Jan 2003 11:02:51 +0000 (11:02 +0000)]
Add optional drop behavior clause to REVOKE command, for SQL conformance.
Currently, only RESTRICT is allowed.

21 years agoClean ip.c.
Peter Eisentraut [Fri, 10 Jan 2003 10:59:08 +0000 (10:59 +0000)]
Clean ip.c.

21 years agoAdjust parser so that 'x NOT IN (subselect)' is converted to
Tom Lane [Thu, 9 Jan 2003 20:50:53 +0000 (20:50 +0000)]
Adjust parser so that 'x NOT IN (subselect)' is converted to
'NOT (x IN (subselect))', that is 'NOT (x = ANY (subselect))',
rather than 'x <> ALL (subselect)' as we formerly did.  This
opens the door to optimizing NOT IN the same way as IN, whereas
there's no hope of optimizing the expression using <>.  Also,
convert 'x <> ALL (subselect)' to the NOT(IN) style, so that
the optimization will be available when processing rules dumped
by older Postgres versions.
initdb forced due to small change in SubLink node representation.

21 years agoAdd missing #include <getopt.h>.
Tom Lane [Thu, 9 Jan 2003 18:27:39 +0000 (18:27 +0000)]
Add missing #include <getopt.h>.

21 years agoMove new typedef AclId into c.h, so as to avoid cluttering namespace
Tom Lane [Thu, 9 Jan 2003 18:00:24 +0000 (18:00 +0000)]
Move new typedef AclId into c.h, so as to avoid cluttering namespace
by having to include miscadmin.h into other header files.

21 years agoFix IPv6 detection and IPv6 to 4 conversion to use only standard
Peter Eisentraut [Thu, 9 Jan 2003 14:35:03 +0000 (14:35 +0000)]
Fix IPv6 detection and IPv6 to 4 conversion to use only standard
interfaces.

21 years agoCall timestamp_in with proper arguments, per gripe from Mark Halliwell.
Tom Lane [Thu, 9 Jan 2003 07:10:07 +0000 (07:10 +0000)]
Call timestamp_in with proper arguments, per gripe from Mark Halliwell.

21 years agoFix indenting of log_min_error_statement.
Bruce Momjian [Thu, 9 Jan 2003 05:19:02 +0000 (05:19 +0000)]
Fix indenting of log_min_error_statement.

21 years agoRemove bit.c/h routines. Not used anymore.
Bruce Momjian [Thu, 9 Jan 2003 01:23:55 +0000 (01:23 +0000)]
Remove bit.c/h routines.  Not used anymore.

21 years agoRepair inconsistent rounding behavior for timestamp, time, interval,
Tom Lane [Thu, 9 Jan 2003 01:06:57 +0000 (01:06 +0000)]
Repair inconsistent rounding behavior for timestamp, time, interval,
per gripe from Csaba Nagy.  There is still potential for platform-specific
behavior for values that are exactly halfway between integers, but at
least we now get the expected answer for all other cases.

21 years agoadd to threads discussion.
Bruce Momjian [Thu, 9 Jan 2003 01:04:35 +0000 (01:04 +0000)]
add to threads discussion.

21 years agoAdd missing pg_proc entry for interval_scale(). The lack of this entry
Tom Lane [Thu, 9 Jan 2003 00:58:41 +0000 (00:58 +0000)]
Add missing pg_proc entry for interval_scale().  The lack of this entry
causes interval rounding not to work as expected in 7.3, for example
SELECT '18:17:15.6'::interval(0) does not round the value.
I did not force initdb, but one is needed to install the added row.

21 years agoRemove 'const' from secure_write, to prevent compiler warning.
Bruce Momjian [Wed, 8 Jan 2003 23:34:22 +0000 (23:34 +0000)]
Remove 'const' from secure_write, to prevent compiler warning.

21 years agoGuard against stopping when numberTuples=0 and counter wraps around.
Tom Lane [Wed, 8 Jan 2003 23:32:29 +0000 (23:32 +0000)]
Guard against stopping when numberTuples=0 and counter wraps around.

21 years agoThe second was that renegotiation was just plain broken. I can't
Bruce Momjian [Wed, 8 Jan 2003 23:18:25 +0000 (23:18 +0000)]
The second was that renegotiation was just plain broken.  I can't
believe I didn't notice this before -- once 64k was sent to/from the
server the client would crash.  Basicly, in 7.3 the server SSL code set
the initial state to "about to renegotiate" without actually starting
the renegotiation.  In addition, the server and client didn't properly
handle the SSL_ERROR_WANT_(READ|WRITE) error.  This is fixed in the
second patch.

Nathan Mueller

21 years agoI was playing around with 7.3.1 and found some more SSL problems. The
Bruce Momjian [Wed, 8 Jan 2003 22:56:58 +0000 (22:56 +0000)]
I was playing around with 7.3.1 and found some more SSL problems.  The
first, that I missed when checking over 7.3.1, was that the client
method was switched to SSLv23 along with the server.  The SSLv23 client
method does SSLv2 by default, but can also understand SSLv3.  In our
situation the SSLv2 backwords compatibility is really only needed on the
server.  This is the first patch.

The last thing is that I found a way for the server to understand SSLv2
HELLO messages (sent by pre-7.3 clients) but then get them to talk
SSLv3.  This is the last one.

Nathan Mueller

21 years agoFix dumping of DEFERRABLE/INITIALLY DEFERRED:
Bruce Momjian [Wed, 8 Jan 2003 22:54:06 +0000 (22:54 +0000)]
Fix dumping of DEFERRABLE/INITIALLY DEFERRED:

> The big problem is that while pg_dump's dump_trigger() looks at
> tginitdeferred and dumps accordingly, pg_get_constraintdef doesn't look
> at tginitdeferred, and therefore doesn't record the requirement as part
> of ALTER TABLE ADD CONSTRAINT.

pg_get_constraintdef should probably be looking at condeferrable and
condeferred in the pg_constraint row it's looking at.  Maybe something
like the attached.

(Added, output only non-default values.)

Stephan Szabo

21 years agoUpdated deferred trigger patch.
Bruce Momjian [Wed, 8 Jan 2003 22:29:23 +0000 (22:29 +0000)]
Updated deferred trigger patch.

Neil Conway

21 years agoThis trivial patch implements disabled, deferred triggers, per my
Bruce Momjian [Wed, 8 Jan 2003 22:28:32 +0000 (22:28 +0000)]
This trivial patch implements disabled, deferred triggers, per my
proposal on -hackers a couple days ago. The 'tgenabled' status of the
trigger is consulted when the trigger is added to the deferred queue
(i.e. when the event occurs that fires the trigger), not when the
deferred event is executed.

No regression tests, as another bug prevents them (the pg_trigger row
for a trigger is only loaded once per session, so any changes to it are
not noticed unless the client disconnects and reconnects).

Neil Conway

21 years agoEnhanced error/help message for pg_controldata.
Bruce Momjian [Wed, 8 Jan 2003 22:26:34 +0000 (22:26 +0000)]
Enhanced error/help message for pg_controldata.

Serguei Mokhov

21 years agoReconsider old decision to try to constant-fold default and constraint
Tom Lane [Wed, 8 Jan 2003 22:06:23 +0000 (22:06 +0000)]
Reconsider old decision to try to constant-fold default and constraint
expressions before they are stored.  This seems like not such a hot idea,
particularly now that the constant-folder will try to inline SQL functions.

21 years agoRepair bug noticed by Deepak Bhole: a shell type should have a dependency
Tom Lane [Wed, 8 Jan 2003 21:40:39 +0000 (21:40 +0000)]
Repair bug noticed by Deepak Bhole: a shell type should have a dependency
on its namespace, so that it will go away if the schema is dropped.

21 years agoFix pgpass to work with PQconnectDB().
Bruce Momjian [Wed, 8 Jan 2003 21:33:27 +0000 (21:33 +0000)]
Fix pgpass to work with PQconnectDB().

Backpatch to 7.3.X.

Report from ljb.

21 years agoFix for bug #866. 7.3 contains new logic for avoiding redundant calls to
Tom Lane [Wed, 8 Jan 2003 19:41:40 +0000 (19:41 +0000)]
Fix for bug #866.  7.3 contains new logic for avoiding redundant calls to
the index AM when we know we are fetching a unique row.  However, this
logic did not consider the possibility that it would be asked to fetch
backwards.  Also fix mark/restore to work correctly in this scenario.

21 years agoOnly set the database name if we find a match in the services file.
Bruce Momjian [Wed, 8 Jan 2003 16:21:53 +0000 (16:21 +0000)]
Only set the database name if we find a match in the services file.

21 years agoCorrection to last patch. As per the DB-API spec, we need to return None
D'Arcy J.M. Cain [Wed, 8 Jan 2003 13:42:16 +0000 (13:42 +0000)]
Correction to last patch.  As per the DB-API spec, we need to return None
here, not -1.

21 years agoAdd:
Bruce Momjian [Wed, 8 Jan 2003 06:14:51 +0000 (06:14 +0000)]
Add:

>  o Allow fastpast to pass values in portable format