OSDN Git Service

pg-rex/syncrep.git
21 years agoClean up script.
Bruce Momjian [Tue, 27 Aug 2002 02:54:39 +0000 (02:54 +0000)]
Clean up script.

21 years agoAdd check_guc utility to compare guc.c and postgresql.conf.sample.
Bruce Momjian [Tue, 27 Aug 2002 02:52:42 +0000 (02:52 +0000)]
Add check_guc utility to compare guc.c and postgresql.conf.sample.

21 years agoFix units in postgresql.conf.
Bruce Momjian [Tue, 27 Aug 2002 02:50:41 +0000 (02:50 +0000)]
Fix units in postgresql.conf.

21 years agoAdd OS X link line example for external functions.
Bruce Momjian [Mon, 26 Aug 2002 23:22:47 +0000 (23:22 +0000)]
Add OS X link line example for external functions.

21 years agoAdd discussion of pre-write pages to WAL.
Bruce Momjian [Mon, 26 Aug 2002 23:14:15 +0000 (23:14 +0000)]
Add discussion of pre-write pages to WAL.

21 years agoAdd:
Bruce Momjian [Mon, 26 Aug 2002 23:13:37 +0000 (23:13 +0000)]
Add:

>   partial page writes [wal]

21 years agoAdd:
Bruce Momjian [Mon, 26 Aug 2002 23:02:21 +0000 (23:02 +0000)]
Add:

> * Add CORRESPONDING BY to UNION/INTERSECT/EXCEPT

21 years agoAdd units to postgresql.conf file.
Bruce Momjian [Mon, 26 Aug 2002 22:23:07 +0000 (22:23 +0000)]
Add units to postgresql.conf file.

21 years agoRemove:
Bruce Momjian [Mon, 26 Aug 2002 22:02:20 +0000 (22:02 +0000)]
Remove:

<  o Add plsh server-side shell language (Peter E)

21 years ago*** empty log message ***
Bruce Momjian [Mon, 26 Aug 2002 22:01:50 +0000 (22:01 +0000)]
*** empty log message ***

21 years agoAdd to selfjoin DELETE todo.
Bruce Momjian [Mon, 26 Aug 2002 21:52:00 +0000 (21:52 +0000)]
Add to selfjoin DELETE todo.

21 years agoAdd:
Bruce Momjian [Mon, 26 Aug 2002 21:35:32 +0000 (21:35 +0000)]
Add:

> * Allow DELETE to handle table aliases for self-joins [delete]

21 years agoAdd delete selfjoin item.
Bruce Momjian [Mon, 26 Aug 2002 21:35:01 +0000 (21:35 +0000)]
Add delete selfjoin item.

21 years agoAdd info on disabled triggers.
Bruce Momjian [Mon, 26 Aug 2002 20:03:10 +0000 (20:03 +0000)]
Add info on disabled triggers.

21 years agoAdd:
Bruce Momjian [Mon, 26 Aug 2002 20:02:45 +0000 (20:02 +0000)]
Add:

> * Allow triggers to be disabled [trigger]

21 years agoIncrease WIDTH_THRESHOLD from 256 to 1K. This addresses recent observation
Tom Lane [Mon, 26 Aug 2002 18:45:57 +0000 (18:45 +0000)]
Increase WIDTH_THRESHOLD from 256 to 1K.  This addresses recent observation
that ANALYZE would not gather any stats for a CHAR(255) column.  I still
think a width threshold is appropriate for the reasons mentioned in the
code, but we can loosen it at least.

21 years agoModify array operations to include array's element type OID in the
Tom Lane [Mon, 26 Aug 2002 17:54:02 +0000 (17:54 +0000)]
Modify array operations to include array's element type OID in the
array header, and to compute sizing and alignment of array elements
the same way normal tuple access operations do --- viz, using the
tupmacs.h macros att_addlength and att_align.  This makes the world
safe for arrays of cstrings or intervals, and should make it much
easier to write array-type-polymorphic functions; as examples see
the cleanups of array_out and contrib/array_iterator.  By Joe Conway
and Tom Lane.

21 years agoAdd 'return' file.
Bruce Momjian [Mon, 26 Aug 2002 17:40:27 +0000 (17:40 +0000)]
Add 'return' file.

21 years agoAdd return tuple count item to TODO.
Bruce Momjian [Mon, 26 Aug 2002 17:40:00 +0000 (17:40 +0000)]
Add return tuple count item to TODO.

21 years agoAdd:
Bruce Momjian [Mon, 26 Aug 2002 17:39:31 +0000 (17:39 +0000)]
Add:

> * Return proper effected tuple count from complex commands [return]

21 years agoAdd anther sequential scan email.
Bruce Momjian [Mon, 26 Aug 2002 17:14:29 +0000 (17:14 +0000)]
Add anther sequential scan email.

21 years agoAdd mention of 1 terrabyte databases.
Bruce Momjian [Mon, 26 Aug 2002 01:05:43 +0000 (01:05 +0000)]
Add mention of 1 terrabyte databases.

21 years ago*** empty log message ***
Bruce Momjian [Mon, 26 Aug 2002 01:04:44 +0000 (01:04 +0000)]
*** empty log message ***

21 years agoAdd raw file discussion to performance TODO.detail.
Bruce Momjian [Mon, 26 Aug 2002 01:04:13 +0000 (01:04 +0000)]
Add raw file discussion to performance TODO.detail.

21 years agoDone:
Bruce Momjian [Mon, 26 Aug 2002 00:52:40 +0000 (00:52 +0000)]
Done:

> * -Disallow TRUNCATE on tables that are involved in referential constraints

21 years agoAdd to tablespaces discussion.
Bruce Momjian [Mon, 26 Aug 2002 00:22:53 +0000 (00:22 +0000)]
Add to tablespaces discussion.

21 years agoAdd to tablespaces file.
Bruce Momjian [Mon, 26 Aug 2002 00:19:33 +0000 (00:19 +0000)]
Add to tablespaces file.

21 years agoAdd Bob Devine's name to the optimizer README.
Bruce Momjian [Sun, 25 Aug 2002 22:39:37 +0000 (22:39 +0000)]
Add Bob Devine's name to the optimizer README.

21 years agoAdd file describing error message format.
Bruce Momjian [Sun, 25 Aug 2002 20:36:14 +0000 (20:36 +0000)]
Add file describing error message format.

21 years agoAdd:
Bruce Momjian [Sun, 25 Aug 2002 20:30:30 +0000 (20:30 +0000)]
Add:

> * Make error messages more consistent [error]

21 years agoClean up comments to be careful about the distinction between variable-
Tom Lane [Sun, 25 Aug 2002 17:20:01 +0000 (17:20 +0000)]
Clean up comments to be careful about the distinction between variable-
width types and varlena types, since with the introduction of CSTRING as
a more-or-less-real type, these concepts aren't identical.  I've tried to
use varlena consistently to denote datatypes with typlen = -1, ie, they
have a length word and are potentially TOASTable; while the term variable
width covers both varlena and cstring (and, perhaps, someday other types
with other rules for computing the actual width).  No code changes in this
commit except for renaming a couple macros.

21 years agoUpdate docs for 7.2.2 release info.
Bruce Momjian [Sun, 25 Aug 2002 14:34:24 +0000 (14:34 +0000)]
Update docs for 7.2.2 release info.

21 years agoUpdate Japanese FAQ, from Jun Kuwamura
Bruce Momjian [Sun, 25 Aug 2002 12:58:44 +0000 (12:58 +0000)]
Update Japanese FAQ, from Jun Kuwamura

21 years agoThe cstring datatype can now be copied, passed around, etc. The typlen
Tom Lane [Sat, 24 Aug 2002 15:00:47 +0000 (15:00 +0000)]
The cstring datatype can now be copied, passed around, etc.  The typlen
value '-2' is used to indicate a variable-width type whose width is
computed as strlen(datum)+1.  Everything that looks at typlen is updated
except for array support, which Joe Conway is working on; at the moment
it wouldn't work to try to create an array of cstring.

21 years agoDone:
Bruce Momjian [Sat, 24 Aug 2002 11:24:58 +0000 (11:24 +0000)]
Done:

> * -SELECT cash_out(2) crashes because of opaque

21 years agoEnhancements to how queries with bind values are stored internally and sent to
Barry Lind [Fri, 23 Aug 2002 20:45:49 +0000 (20:45 +0000)]
Enhancements to how queries with bind values are stored internally and sent to
the server.  Previously we allocated a new String object for the entire final
query we were sending to the database.  If you had a big query, or especially
if you had large bind values you ended up with essentially two copies in memory.
This change will reuse the existing objects and therefore should take 1/2 the
memory it does today for a given query.  This restructuring will also allow
in the future the ability to stream bytea data to the server instead of the current approach of pulling it all into memory.
I also fixed a test that was failing on a 7.2 database.
Also renamed some internal variables and some minor cleanup.

 Modified Files:
  jdbc/org/postgresql/core/QueryExecutor.java
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Connection.java
  jdbc/org/postgresql/jdbc1/AbstractJdbc1Statement.java
  jdbc/org/postgresql/jdbc2/AbstractJdbc2ResultSet.java
  jdbc/org/postgresql/jdbc2/AbstractJdbc2Statement.java
  jdbc/org/postgresql/test/jdbc2/DatabaseMetaDataTest.java

21 years agoFurther cleanup around the edges of OPAQUE/pseudotype changes. Correct
Tom Lane [Fri, 23 Aug 2002 16:41:38 +0000 (16:41 +0000)]
Further cleanup around the edges of OPAQUE/pseudotype changes.  Correct
the declarations of some index access method support functions.  Support
SQL functions returning VOID.

21 years agoAdapt for SRF(Set Returning Function).
Tatsuo Ishii [Fri, 23 Aug 2002 08:19:49 +0000 (08:19 +0000)]
Adapt for SRF(Set Returning Function).

21 years agoNot sure how I fixed it the first time, but here's a fix for another
Bruce Momjian [Fri, 23 Aug 2002 04:27:19 +0000 (04:27 +0000)]
Not sure how I fixed it the first time, but here's a fix for another
instance in which the docs mention that 'NAMEDATALEN == 32".

Neil Conway

21 years agoAdd space.k>
Bruce Momjian [Fri, 23 Aug 2002 03:10:44 +0000 (03:10 +0000)]
Add space.k>

21 years agoAdd:
Bruce Momjian [Fri, 23 Aug 2002 03:10:15 +0000 (03:10 +0000)]
Add:

> * Allow xlog directory location to be specified during initdb, perhaps
>   using symlinks

21 years agoPatch resolve ERROR problem for non-goog query_txt.
Bruce Momjian [Fri, 23 Aug 2002 02:56:36 +0000 (02:56 +0000)]
Patch resolve ERROR problem for non-goog query_txt.

Teodor Sigaev

21 years agoThis patch updates the CREATE OPERATOR sgml docs for the new default
Bruce Momjian [Fri, 23 Aug 2002 02:54:18 +0000 (02:54 +0000)]
This patch updates the CREATE OPERATOR sgml docs for the new default
settings of NAMEDATALEN. I looked through the docs for other
references to NAMEDATALEN, but this is the only one I could find.

Neil Conway

21 years agoUpdate Russian FAQ.
Bruce Momjian [Fri, 23 Aug 2002 02:53:20 +0000 (02:53 +0000)]
Update Russian FAQ.

21 years agoAdd:
Bruce Momjian [Fri, 23 Aug 2002 02:46:39 +0000 (02:46 +0000)]
Add:

> * Add GUC parameter to print queries that generate errors

21 years agoVery minor copy-editing.
Tom Lane [Fri, 23 Aug 2002 01:28:17 +0000 (01:28 +0000)]
Very minor copy-editing.

21 years agoAdd note that explains that \df omits functions it thinks are I/O functions.
Tom Lane [Fri, 23 Aug 2002 01:27:44 +0000 (01:27 +0000)]
Add note that explains that \df omits functions it thinks are I/O functions.

21 years agoUpdate reference to value of NAMEDATALEN. Someone needs to troll the
Tom Lane [Fri, 23 Aug 2002 00:33:24 +0000 (00:33 +0000)]
Update reference to value of NAMEDATALEN.  Someone needs to troll the
docs and find the other obsolete statements that no doubt lurk.

21 years agoUpdate contrib regression tests for OPAQUE datatype changes.
Tom Lane [Fri, 23 Aug 2002 00:04:52 +0000 (00:04 +0000)]
Update contrib regression tests for OPAQUE datatype changes.

21 years agoSince the plpgsql test has been part of the standard regression tests
Tom Lane [Thu, 22 Aug 2002 23:33:43 +0000 (23:33 +0000)]
Since the plpgsql test has been part of the standard regression tests
for a good long while, I see no reason to maintain this original copy.

21 years agoTranspose info from src/pl/plpgsql/test/README into standard regression test file.
Tom Lane [Thu, 22 Aug 2002 23:31:48 +0000 (23:31 +0000)]
Transpose info from src/pl/plpgsql/test/README into standard regression test file.

21 years agoMark 7.2.2 as newest release.
Bruce Momjian [Thu, 22 Aug 2002 23:20:52 +0000 (23:20 +0000)]
Mark 7.2.2 as newest release.

21 years agoUpdate FAQ.
Bruce Momjian [Thu, 22 Aug 2002 23:20:25 +0000 (23:20 +0000)]
Update FAQ.

21 years agoRemove all traces of the ODBC driver, which is now on GBorg as the psqlodbc
Marc G. Fournier [Thu, 22 Aug 2002 22:43:14 +0000 (22:43 +0000)]
Remove all traces of the ODBC driver, which is now on GBorg as the psqlodbc
project ...

21 years agoFix bit rot in pg_dump's ability to dump from 7.2 and 7.1 servers.
Tom Lane [Thu, 22 Aug 2002 21:35:50 +0000 (21:35 +0000)]
Fix bit rot in pg_dump's ability to dump from 7.2 and 7.1 servers.

21 years agoFixed-size buffer in dumpClasses is not big enough anymore given the
Tom Lane [Thu, 22 Aug 2002 21:29:34 +0000 (21:29 +0000)]
Fixed-size buffer in dumpClasses is not big enough anymore given the
addition of a column list clause to the COPY command.  Spotted by
Martin Renters.

21 years agoImprove wording of upgrade section.
Bruce Momjian [Thu, 22 Aug 2002 15:31:07 +0000 (15:31 +0000)]
Improve wording of upgrade section.

21 years agoCode review for recent TRUNCATE changes. Tighten relation-kind check,
Tom Lane [Thu, 22 Aug 2002 14:23:36 +0000 (14:23 +0000)]
Code review for recent TRUNCATE changes.  Tighten relation-kind check,
tighten foreign-key check (a self-reference should not prevent TRUNCATE),
improve error message, cause a relation's TOAST table to be truncated
along with the relation.

21 years agoCommit updated repeat() patch, from Neil Conway
Bruce Momjian [Thu, 22 Aug 2002 05:05:19 +0000 (05:05 +0000)]
Commit updated repeat() patch, from  Neil Conway

21 years agoFix for documention:
Bruce Momjian [Thu, 22 Aug 2002 04:56:44 +0000 (04:56 +0000)]
Fix for documention:

>>" It's also possible to select no escape character by writing ESCAPE ''.
>>In this case there is no way to turn off the special meaning of
>>underscore and percent signs in the pattern."

Joe Conway

21 years agoThis patch should fix the problem. Doesn't include my previous patch
Bruce Momjian [Thu, 22 Aug 2002 04:55:05 +0000 (04:55 +0000)]
This patch should fix the problem. Doesn't include my previous patch
for repeat(). Again, somewhat off-the-cuff, so I might have missed
something...

test=# select lpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');
ERROR:  Requested length too large
test=# select rpad('xxxxx',1431655765,'yyyyyyyyyyyyyyyy');
ERROR:  Requested length too large

(That's on a Unicode DB, haven't tested other encodings but AFAICT
this fix should still work.)

Neil Conway

21 years agorepeat() fix:
Bruce Momjian [Thu, 22 Aug 2002 04:54:20 +0000 (04:54 +0000)]
repeat() fix:

> Neil Conway <neilc@samurai.com> writes:
> > +   /* Check for integer overflow */
> > +   if (tlen / slen != count)
> > +           elog(ERROR, "Requested buffer is too large.");
>
> What about slen == 0?

Good point -- that wouldn't cause incorrect results or a security
problem, but it would reject input that we should really accept.

Revised patch is attached.

Neil Conway

21 years ago> > I had great difficulty in finding how to change the search path, so here
Bruce Momjian [Thu, 22 Aug 2002 04:52:17 +0000 (04:52 +0000)]
> > I had great difficulty in finding how to change the search path, so here
> > is a patch to add some cross-referencing.

Oliver Elphick

21 years ago# Disallow TRUNCATE on tables that are involved in referential
Bruce Momjian [Thu, 22 Aug 2002 04:51:06 +0000 (04:51 +0000)]
# Disallow TRUNCATE on tables that are involved in referential
constraints

The issue with finding and removing foreign key constraints is no longer
an issue, so please apply the attached.

It does NOT check for rules or on delete triggers (old style foreign
keys) as those are difficult to deal with (remove, truncate, re-add).

Rod Taylor

21 years agoAdd:
Bruce Momjian [Thu, 22 Aug 2002 04:47:31 +0000 (04:47 +0000)]
Add:

> * Allow bytea to handle LIKE with non-TEXT patterns

21 years agoAs suggested by Tom, this patch restricts the right-hand argument of
Bruce Momjian [Thu, 22 Aug 2002 04:45:11 +0000 (04:45 +0000)]
As suggested by Tom, this patch restricts the right-hand argument of
bytealike to TEXT.

This leaves like_escape_bytea() without anything to do, but I left it in
place in anticipation of the eventual bytea pattern selectivity
functions. If there is agreement that this would be the best long term
solution, I'll take it as a TODO for 7.4.

Joe Conway

21 years agoOops, we got duplicate oids from patches again; pick a unique one.
Bruce Momjian [Thu, 22 Aug 2002 04:41:06 +0000 (04:41 +0000)]
Oops, we got duplicate oids from patches again;  pick a unique one.

21 years agoAdd:
Bruce Momjian [Thu, 22 Aug 2002 03:24:01 +0000 (03:24 +0000)]
Add:

replace(string, from, to)
   -- replaces all occurrences of "from" in "string" to "to"
split(string, fldsep, column)
   -- splits "string" on "fldsep" and returns "column" number piece
to_hex(int32_num) & to_hex(int64_num)
   -- takes integer number and returns as hex string

Joe Conway

21 years agoRemove libpq++ SGML documentation.
Bruce Momjian [Thu, 22 Aug 2002 03:17:38 +0000 (03:17 +0000)]
Remove libpq++ SGML documentation.

21 years agoRemove erroneous character from Makefile due to editor error.
Bruce Momjian [Thu, 22 Aug 2002 02:18:45 +0000 (02:18 +0000)]
Remove erroneous character from Makefile due to editor error.

21 years agoOkay, libpq++ is moved to GBorg, and all traces of it have been removed
Marc G. Fournier [Thu, 22 Aug 2002 00:15:14 +0000 (00:15 +0000)]
Okay, libpq++ is moved to GBorg, and all traces of it have been removed
from the core repository ... I haven't *moved* the libpq++ files out of the
tree, mainly as we want to keep them in place for past branches ...

Peter, I think I've covered all the files I need, and re-ran autoconf to make
sure the configure file is in place properly ...

21 years agoAdd a bunch of pseudo-types to replace the behavior formerly associated
Tom Lane [Thu, 22 Aug 2002 00:01:51 +0000 (00:01 +0000)]
Add a bunch of pseudo-types to replace the behavior formerly associated
with OPAQUE, as per recent pghackers discussion.  I still want to do some
more work on the 'cstring' pseudo-type, but I'm going to commit the bulk
of the changes now before the tree starts shifting under me ...

21 years agoAdd --help and --version options. Message cleanup. NLS.
Peter Eisentraut [Wed, 21 Aug 2002 22:24:34 +0000 (22:24 +0000)]
Add --help and --version options.  Message cleanup.  NLS.

21 years agoChanges made so new conversion Makefiles will build out of the source tree.
Bruce Momjian [Wed, 21 Aug 2002 21:33:55 +0000 (21:33 +0000)]
Changes made so new conversion Makefiles will build out of the source tree.

21 years agoMove PO files into subdirectories separate from the source code.
Peter Eisentraut [Wed, 21 Aug 2002 20:42:27 +0000 (20:42 +0000)]
Move PO files into subdirectories separate from the source code.

21 years agoAdd sys/types.h to common includes.
Peter Eisentraut [Wed, 21 Aug 2002 17:20:58 +0000 (17:20 +0000)]
Add sys/types.h to common includes.

21 years agoFix bad markup detected by Alvaro Herrera
Bruce Momjian [Wed, 21 Aug 2002 16:08:19 +0000 (16:08 +0000)]
Fix bad markup detected by Alvaro Herrera

21 years agoFix markup, Alvaro Herrera.
Bruce Momjian [Wed, 21 Aug 2002 05:25:49 +0000 (05:25 +0000)]
Fix markup,  Alvaro Herrera.

21 years agoUse a non-duplicate OID for pg_lock_status.
Tom Lane [Tue, 20 Aug 2002 19:23:07 +0000 (19:23 +0000)]
Use a non-duplicate OID for pg_lock_status.

21 years agoEnable large file support.
Peter Eisentraut [Tue, 20 Aug 2002 17:54:45 +0000 (17:54 +0000)]
Enable large file support.

Use off_t and size_t in pg_dump to handle file offset arithmetic correctly.

21 years agoIncrease buffer size in cash_words(). Pure paranoia; I don't think the
Tom Lane [Tue, 20 Aug 2002 16:46:29 +0000 (16:46 +0000)]
Increase buffer size in cash_words().  Pure paranoia; I don't think the
code is broken, but any small change in the output format might overrun
the buffer with the old size.

21 years agoUpdate expected files for recent changes of domain tests.
Tom Lane [Tue, 20 Aug 2002 16:37:03 +0000 (16:37 +0000)]
Update expected files for recent changes of domain tests.
From Rod Taylor.

21 years agoThis patch improves the behavior of FOUND in PL/PgSQL. In Oracle,
Bruce Momjian [Tue, 20 Aug 2002 05:28:24 +0000 (05:28 +0000)]
This patch improves the behavior of FOUND in PL/PgSQL. In Oracle,
FOUND is set whenever a SELECT INTO returns > 0 rows, *or* when an
INSERT, UPDATE, or DELETE affects > 0 rows. We implemented the first
part of this behavior, but not the second.

I also improved the documentation on the various situations in which
FOUND can be set (excluding inside FOR loops, which I still need to
think about), and added some regression tests for this behavior.

Neil Conway

21 years agoCleanup.
Bruce Momjian [Tue, 20 Aug 2002 05:12:48 +0000 (05:12 +0000)]
Cleanup.

21 years ago- test for indexes on a domain
Bruce Momjian [Tue, 20 Aug 2002 04:48:36 +0000 (04:48 +0000)]
- test for indexes on a domain
- don't fail type-sanity checks if a domain exists

Rod Taylor

21 years agoAllow pg_statistics to be reset by calling pg_stat_reset().
Bruce Momjian [Tue, 20 Aug 2002 04:47:52 +0000 (04:47 +0000)]
Allow pg_statistics to be reset by calling pg_stat_reset().

Christopher Kings-Lynne

21 years agoAdd current_database().
Bruce Momjian [Tue, 20 Aug 2002 04:46:00 +0000 (04:46 +0000)]
Add current_database().

> Quick system function to pull out the current database.
>
> I've used this a number of times to allow stored procedures to find out
> where they are.  Especially useful for those that do logging or hit a
> remote server.
>
> It's called current_database() to match with current_user().

It's also a necessity for an informational schema.  The catalog
(database) name is required in a number of places.

Rod Taylor

21 years agoRemoved code that is no longer used and has been commented out
Barry Lind [Tue, 20 Aug 2002 04:26:02 +0000 (04:26 +0000)]
Removed code that is no longer used and has been commented out
for the last two releases.

 Modified Files:
  jdbc/org/postgresql/Driver.java.in
  jdbc/org/postgresql/PG_Stream.java
 Removed Files:
  jdbc/org/postgresql/core/BytePoolDim1.java
  jdbc/org/postgresql/core/BytePoolDim2.java
  jdbc/org/postgresql/core/MemoryPool.java
  jdbc/org/postgresql/core/ObjectPool.java
  jdbc/org/postgresql/core/SimpleObjectPool.java

21 years agoAdd mention of -F not supported by pg_dumpall. Code already disallows it.
Bruce Momjian [Tue, 20 Aug 2002 02:20:48 +0000 (02:20 +0000)]
Add mention of -F not supported by pg_dumpall.  Code already disallows it.

21 years agoDone by Rod:
Bruce Momjian [Tue, 20 Aug 2002 02:16:38 +0000 (02:16 +0000)]
Done by Rod:

> * -Make other constraints clearer in dump file

21 years agoFix pg_dump to dump serial columns as serials. Per pghackers discussion,
Tom Lane [Mon, 19 Aug 2002 19:33:36 +0000 (19:33 +0000)]
Fix pg_dump to dump serial columns as serials.  Per pghackers discussion,
cause SERIAL column declaration not to imply UNIQUE, so that this can be
done without creating an extra index.

21 years agoRemove Ident nodetype in favor of using String nodes; this fixes some
Tom Lane [Mon, 19 Aug 2002 15:08:47 +0000 (15:08 +0000)]
Remove Ident nodetype in favor of using String nodes; this fixes some
latent wrong-struct-type bugs and makes the coding style more uniform,
since the majority of places working with lists of column names were
already using Strings not Idents.  While at it, remove vestigial
support for Stream node type, and otherwise-unreferenced nodes.h entries
for T_TupleCount and T_BaseNode.
NB: full recompile is recommended due to changes of Node type numbers.
This shouldn't force an initdb though.

21 years agoFix bug in pg_convert() per report from MaC.Yui.
Tatsuo Ishii [Mon, 19 Aug 2002 04:08:08 +0000 (04:08 +0000)]
Fix bug in pg_convert() per report from MaC.Yui.
It pfree() wrong pointer.

21 years agoFix broken tag.
Tatsuo Ishii [Mon, 19 Aug 2002 04:05:00 +0000 (04:05 +0000)]
Fix broken tag.

21 years agoClean up leftover bugs from recent COPY feature patch --- missed
Tom Lane [Mon, 19 Aug 2002 00:40:15 +0000 (00:40 +0000)]
Clean up leftover bugs from recent COPY feature patch --- missed
required changes to copyfuncs/equalfuncs.

21 years agoAdd missing copyfuncs/equalfuncs entries, including T_Null which has
Tom Lane [Mon, 19 Aug 2002 00:11:53 +0000 (00:11 +0000)]
Add missing copyfuncs/equalfuncs entries, including T_Null which has
been missing forever; surprising it wasn't noticed before.  The other
additions are, um, sloppiness in certain recent feature additions.

21 years agoImprove obsolete comment.
Tom Lane [Mon, 19 Aug 2002 00:10:03 +0000 (00:10 +0000)]
Improve obsolete comment.

21 years agoFix small copy-and-pasteo.
Tom Lane [Sun, 18 Aug 2002 21:05:32 +0000 (21:05 +0000)]
Fix small copy-and-pasteo.

21 years agoFix potential buffer overrun in cube_out(), per report from
Tom Lane [Sun, 18 Aug 2002 20:15:52 +0000 (20:15 +0000)]
Fix potential buffer overrun in cube_out(), per report from
Bruno Wolff.

21 years agoRemove optimization whereby parser would make only one sort-list entry
Tom Lane [Sun, 18 Aug 2002 18:46:15 +0000 (18:46 +0000)]
Remove optimization whereby parser would make only one sort-list entry
when two equal() targetlist items were to be added to an ORDER BY or
DISTINCT list.  Although indeed this would make sorting fractionally
faster by sometimes saving a comparison, it confuses the heck out of
later stages of processing, because it makes it look like the user
wrote DISTINCT ON rather than DISTINCT.  Bug reported by joe@piscitella.com.