]> granicus.if.org Git - postgresql/log
postgresql
25 years agoMajor psql overhaul by Peter Eisentraut.
Bruce Momjian [Thu, 4 Nov 1999 21:56:02 +0000 (21:56 +0000)]
Major psql overhaul by Peter Eisentraut.

25 years agoautoconf
Bruce Momjian [Thu, 4 Nov 1999 21:48:36 +0000 (21:48 +0000)]
autoconf

25 years agoImprove getopt and readline support, from Peter Eisentraut.
Bruce Momjian [Thu, 4 Nov 1999 21:46:40 +0000 (21:46 +0000)]
Improve getopt and readline support, from Peter Eisentraut.

25 years agoUpdate bsdi for bsd/os 4.1
Bruce Momjian [Thu, 4 Nov 1999 16:22:23 +0000 (16:22 +0000)]
Update bsdi for bsd/os 4.1

25 years agoMake it possible to execute crashed CREATE/DROP commands again.
Hiroshi Inoue [Thu, 4 Nov 1999 08:01:09 +0000 (08:01 +0000)]
Make it possible to execute crashed CREATE/DROP commands again.
Now indexes of pg_class and pg_type are unique indexes
and guarantee the uniqueness of correponding attributes.
heap_create() was changed to take another boolean parameter
which allows to postpone the creation of disk file.
The name of rd_nonameunlinked was changed to rd_unlinked.
It is used generally(not only for noname relations) now.
Requires initdb.

25 years ago*** empty log message ***
Michael Meskes [Tue, 2 Nov 1999 19:03:34 +0000 (19:03 +0000)]
*** empty log message ***

25 years ago*** empty log message ***
Michael Meskes [Tue, 2 Nov 1999 12:11:53 +0000 (12:11 +0000)]
*** empty log message ***

25 years agoupdate for pgaccess
Bruce Momjian [Tue, 2 Nov 1999 00:05:29 +0000 (00:05 +0000)]
update for pgaccess

25 years agoautoconf
Bruce Momjian [Mon, 1 Nov 1999 23:59:45 +0000 (23:59 +0000)]
autoconf

25 years agoUpdate pgaccess
Bruce Momjian [Mon, 1 Nov 1999 23:55:00 +0000 (23:55 +0000)]
Update pgaccess

25 years agoautoconf
Bruce Momjian [Mon, 1 Nov 1999 23:51:55 +0000 (23:51 +0000)]
autoconf

25 years agoUpdate for pgaccess 0.98.
Bruce Momjian [Mon, 1 Nov 1999 23:48:27 +0000 (23:48 +0000)]
Update for pgaccess 0.98.

25 years agoUpdate for 6.5.3.
Bruce Momjian [Mon, 1 Nov 1999 16:20:46 +0000 (16:20 +0000)]
Update for 6.5.3.

25 years agoEliminate RewritePreprocessQuery, which was taking an
Tom Lane [Mon, 1 Nov 1999 05:18:31 +0000 (05:18 +0000)]
Eliminate RewritePreprocessQuery, which was taking an
unreasonable amount of time to clean up after a vanished parser problem.
Don't call fireRIRonSubselect when we know there are no subselects,
either.

25 years agoEliminate some unbelievably cheesy code in _copyConst().
Tom Lane [Mon, 1 Nov 1999 05:15:13 +0000 (05:15 +0000)]
Eliminate some unbelievably cheesy code in _copyConst().
Apparently, back in the dim reaches of prehistory, the parser couldn't
be trusted to label Const nodes with the correct constbyval value ...
and someone preferred to patch around this in copyObject rather than
fix the problem at the source.  The problem is long gone, but the hack
lingered on.  Until now.

25 years agoInline check for full buffer in appendStringInfoChar.
Tom Lane [Mon, 1 Nov 1999 05:10:32 +0000 (05:10 +0000)]
Inline check for full buffer in appendStringInfoChar.

25 years agoDon't call ExecOpenIndices if pg_class relhasindex shows there are no
Tom Lane [Mon, 1 Nov 1999 05:09:18 +0000 (05:09 +0000)]
Don't call ExecOpenIndices if pg_class relhasindex shows there are no
indexes to open.  Avoid unnecessary work in ExecCheckPerm, too.

25 years agoEliminate local inefficiencies in updateTargetListEntry, make_var, and
Tom Lane [Mon, 1 Nov 1999 05:06:21 +0000 (05:06 +0000)]
Eliminate local inefficiencies in updateTargetListEntry, make_var, and
make_const --- don't repeat cache searches that aren't needed.

25 years agoformatting cleanup
Bruce Momjian [Mon, 1 Nov 1999 04:00:57 +0000 (04:00 +0000)]
formatting cleanup

25 years agoUpdate version for 7.0.
Bruce Momjian [Mon, 1 Nov 1999 02:33:32 +0000 (02:33 +0000)]
Update version for 7.0.

25 years agoAllow indexes on system catalogs for use in cache code.
Bruce Momjian [Mon, 1 Nov 1999 02:29:27 +0000 (02:29 +0000)]
Allow indexes on system catalogs for use in cache code.

Thanks to Hiroshi

25 years agoautoconf
Bruce Momjian [Sun, 31 Oct 1999 19:52:28 +0000 (19:52 +0000)]
autoconf

25 years agoUpdate to pgaccess 0.98.
Bruce Momjian [Sun, 31 Oct 1999 19:50:48 +0000 (19:50 +0000)]
Update to pgaccess 0.98.

25 years agoRule dumper failed to affix '*' to inheritable RTEs.
Tom Lane [Sun, 31 Oct 1999 18:57:42 +0000 (18:57 +0000)]
Rule dumper failed to affix '*' to inheritable RTEs.

25 years agoUpdate TODO list.
Bruce Momjian [Sun, 31 Oct 1999 13:19:42 +0000 (13:19 +0000)]
Update TODO list.

25 years agoUpdate TODO list.
Bruce Momjian [Sun, 31 Oct 1999 13:10:47 +0000 (13:10 +0000)]
Update TODO list.

25 years agoUpdate for 0.98 pgaccess
Bruce Momjian [Sun, 31 Oct 1999 12:33:44 +0000 (12:33 +0000)]
Update for 0.98 pgaccess

25 years agoUpdate for 0.98.
Bruce Momjian [Sun, 31 Oct 1999 12:23:41 +0000 (12:23 +0000)]
Update for 0.98.

25 years agoautoconf
Bruce Momjian [Sun, 31 Oct 1999 11:51:29 +0000 (11:51 +0000)]
autoconf

25 years agoUpdate pgaccess makefile for 0.98
Bruce Momjian [Sun, 31 Oct 1999 11:50:03 +0000 (11:50 +0000)]
Update pgaccess makefile for 0.98

25 years agoAvoid duplicate ExecTypeFromTL() call in ExecInitJunkFilter() by passing
Tom Lane [Sat, 30 Oct 1999 23:13:30 +0000 (23:13 +0000)]
Avoid duplicate ExecTypeFromTL() call in ExecInitJunkFilter() by passing
in the TupleDesc that the caller already has (for call from ExecMain) or
can make just as easily as ExecInitJunkFilter() can (for call from
ExecAppend).  Also, don't bother to build a junk filter for an INSERT
operation that doesn't actually need one, which is the normal case.

25 years agoRearrange heap_beginscan to eliminate extra call on
Tom Lane [Sat, 30 Oct 1999 23:10:21 +0000 (23:10 +0000)]
Rearrange heap_beginscan to eliminate extra call on
mdnblocks.  Bad enough it does an lseek, but to do it twice for no
reason...

25 years agoSkip invoking set_uppernode_references() for a RESULT node
Tom Lane [Sat, 30 Oct 1999 23:07:55 +0000 (23:07 +0000)]
Skip invoking set_uppernode_references() for a RESULT node
that has no subplan --- saves a material amount of time for a simple
INSERT ... VALUES query.

25 years agoRewrite preprocess_targetlist() to reduce overhead for simple INSERTs.
Tom Lane [Sat, 30 Oct 1999 23:06:32 +0000 (23:06 +0000)]
Rewrite preprocess_targetlist() to reduce overhead for simple INSERTs.
In particular, don't bother to look up type information for attributes
where we're not actually going to use it, and avoid copying entire tlist
structure when it's not necessary.

25 years agoUpdate pgaccess for 0.98.
Bruce Momjian [Sat, 30 Oct 1999 21:21:50 +0000 (21:21 +0000)]
Update pgaccess for 0.98.

25 years agoFurther performance improvements in sorting: reduce number of comparisons
Tom Lane [Sat, 30 Oct 1999 17:27:15 +0000 (17:27 +0000)]
Further performance improvements in sorting: reduce number of comparisons
during initial run formation by keeping both current run and next-run
tuples in the same heap (yup, Knuth is smarter than I am).  And, during
merge passes, make use of available sort memory to load multiple tuples
from any one input 'tape' at a time, thereby improving locality of
access to the temp file.

25 years agoAdd pgaccess Makefile.
Bruce Momjian [Sat, 30 Oct 1999 16:39:48 +0000 (16:39 +0000)]
Add pgaccess Makefile.

25 years agoAdd 0.98.
Bruce Momjian [Sat, 30 Oct 1999 16:18:54 +0000 (16:18 +0000)]
Add 0.98.

25 years agoRemove 0.96 for update to 0.98
Bruce Momjian [Sat, 30 Oct 1999 16:01:19 +0000 (16:01 +0000)]
Remove 0.96 for update to 0.98

25 years agoOoops ... 6.5 coding wasn't quite right anymore. Should learn
Tom Lane [Sat, 30 Oct 1999 02:35:14 +0000 (02:35 +0000)]
Ooops ... 6.5 coding wasn't quite right anymore.  Should learn
never to commit without running regress tests...

25 years agoPut back code in nodeAgg to generate a dummy all-nulls input tuple
Tom Lane [Sat, 30 Oct 1999 01:18:16 +0000 (01:18 +0000)]
Put back code in nodeAgg to generate a dummy all-nulls input tuple
before calling execProject, when the outerPlan has returned zero tuples.
I took this out under the mistaken impression that the input tuple
couldn't be referenced by execProject if we weren't in GROUP BY mode.
But it can, if we're in an UPDATE or DELETE...

25 years agoMake USING in COPY optional.
Bruce Momjian [Fri, 29 Oct 1999 23:52:22 +0000 (23:52 +0000)]
Make USING in COPY optional.

25 years agoFix wording.
Bruce Momjian [Fri, 29 Oct 1999 23:44:42 +0000 (23:44 +0000)]
Fix wording.

25 years ago*** empty log message ***
Michael Meskes [Fri, 29 Oct 1999 19:39:31 +0000 (19:39 +0000)]
*** empty log message ***

25 years agoUpdate TODO list.
Bruce Momjian [Wed, 27 Oct 1999 18:28:59 +0000 (18:28 +0000)]
Update TODO list.

25 years agoFix psql quote missing.
Bruce Momjian [Wed, 27 Oct 1999 16:33:45 +0000 (16:33 +0000)]
Fix psql quote missing.

25 years ago*** empty log message ***
Michael Meskes [Wed, 27 Oct 1999 14:36:09 +0000 (14:36 +0000)]
*** empty log message ***

25 years agoAdd mention of NULL for COMMENT.
Bruce Momjian [Tue, 26 Oct 1999 16:36:28 +0000 (16:36 +0000)]
Add mention of NULL for COMMENT.

25 years agoFix compile after COMMENT problem.
Bruce Momjian [Tue, 26 Oct 1999 16:32:46 +0000 (16:32 +0000)]
Fix compile after COMMENT problem.

25 years agoupdate
Bruce Momjian [Tue, 26 Oct 1999 05:16:00 +0000 (05:16 +0000)]
update

25 years agoUpdate TODO list.
Bruce Momjian [Tue, 26 Oct 1999 04:59:12 +0000 (04:59 +0000)]
Update TODO list.

25 years agoThese are two minor corrections to libpq's PQsetNoticeProcessor function.
Bruce Momjian [Tue, 26 Oct 1999 04:49:00 +0000 (04:49 +0000)]
These are two minor corrections to libpq's PQsetNoticeProcessor function.

One, it now returns the previous hook. That way people don't have to dig
around in libpq-int.h for that information anymore. It previously
returned void, so there should be no incompatibilities.

Second, you cannot set the callback to NULL anymore. (Of course you can
still call it with NULL just to get the current hook.) The way libpq uses
the callback pointer, having a NULL there wasn't very healthy.

Peter Eisentraut

25 years agoimprove manual page
Bruce Momjian [Tue, 26 Oct 1999 04:40:58 +0000 (04:40 +0000)]
improve manual page

25 years agoUpdate TODO list.
Bruce Momjian [Tue, 26 Oct 1999 04:21:59 +0000 (04:21 +0000)]
Update TODO list.

25 years agoUpdate TODO list.
Bruce Momjian [Tue, 26 Oct 1999 04:12:34 +0000 (04:12 +0000)]
Update TODO list.

25 years agoMore comments.
Bruce Momjian [Tue, 26 Oct 1999 03:53:35 +0000 (03:53 +0000)]
More comments.

25 years agoAdd documentation for comment.
Bruce Momjian [Tue, 26 Oct 1999 03:48:58 +0000 (03:48 +0000)]
Add documentation for comment.

25 years agoHello.
Bruce Momjian [Tue, 26 Oct 1999 03:12:39 +0000 (03:12 +0000)]
Hello.

The following patch extends the COMMENT ON functionality to the
rest of the database objects beyond just tables, columns, and views. The
grammer of the COMMENT ON statement now looks like:

COMMENT ON [
  [ DATABASE | INDEX | RULE | SEQUENCE | TABLE | TYPE | VIEW ] <objname>
|

  COLUMN <relation>.<attribute> |
  AGGREGATE <aggname> <aggtype> |
  FUNCTION <funcname> (arg1, arg2, ...) |
  OPERATOR <op> (leftoperand_typ rightoperand_typ) |
  TRIGGER <triggername> ON relname>

Mike Mascari
(mascarim@yahoo.com)

25 years agoStandardize on MAXPGPATH as the size of a file pathname buffer,
Tom Lane [Mon, 25 Oct 1999 03:08:03 +0000 (03:08 +0000)]
Standardize on MAXPGPATH as the size of a file pathname buffer,
eliminating some wildly inconsistent coding in various parts of the
system.  I set MAXPGPATH = 1024 in config.h.in.  If anyone is really
convinced that there ought to be a configure-time test to set the
value, go right ahead ... but I think it's a waste of time.

25 years ago'extern char *pg_pathname' in these files fails to agree with
Tom Lane [Mon, 25 Oct 1999 02:31:47 +0000 (02:31 +0000)]
'extern char *pg_pathname' in these files fails to agree with
postgres.c's declaration of 'char pg_pathname[...]'.  I dunno when these
ports were last used, but they are sure broken now...

25 years agobug fix for large objects (free)
Byron Nikolaidis [Mon, 25 Oct 1999 01:11:44 +0000 (01:11 +0000)]
bug fix for large objects (free)

25 years agoAdd a notion of a 'catalog version number' that can indicate
Tom Lane [Sun, 24 Oct 1999 20:42:27 +0000 (20:42 +0000)]
Add a notion of a 'catalog version number' that can indicate
when an initdb-forcing change has been applied within a development cycle.
PG_VERSION serves this purpose for official releases, but we can't bump
the PG_VERSION number every time we make a change to the catalogs during
development.  Instead, increase the catalog version number to warn other
developers that you've made an incompatible change.  See my mail to
pghackers for more info.

25 years agoAdd comment explaining that this table doesn't do what you
Tom Lane [Sun, 24 Oct 1999 19:22:37 +0000 (19:22 +0000)]
Add comment explaining that this table doesn't do what you
might think ... in fact doesn't do much of anything at the moment ...

25 years agoRemove gratuitous redefinitions of BLCKSZ.
Tom Lane [Sat, 23 Oct 1999 03:28:58 +0000 (03:28 +0000)]
Remove gratuitous redefinitions of BLCKSZ.

25 years agoLook Ma, no MAX_PARSE_BUFFER! (At least not in the backend.
Tom Lane [Sat, 23 Oct 1999 03:13:33 +0000 (03:13 +0000)]
Look Ma, no MAX_PARSE_BUFFER!  (At least not in the backend.
pg_dump and interfaces/odbc still need some work.)

25 years agoBackup psql banner change.
Bruce Momjian [Sat, 23 Oct 1999 01:31:32 +0000 (01:31 +0000)]
Backup psql banner change.

25 years agoRemove fixed-size literal buffer from ecpg's lexer (same
Tom Lane [Fri, 22 Oct 1999 23:14:50 +0000 (23:14 +0000)]
Remove fixed-size literal buffer from ecpg's lexer (same
fix recently applied to backend's lexer).  I see that YY_USES_REJECT
still gets defined for this lexer, which means it's going to have trouble
parsing really long tokens.  Not sure if it's worth doing anything about
that or not; I don't have the interest right now to understand why
ecpg's additions to the syntax cause this problem...

25 years agoRemove libpq++ todo item.
Bruce Momjian [Fri, 22 Oct 1999 19:13:39 +0000 (19:13 +0000)]
Remove libpq++ todo item.

25 years agoThis *should* fix the large binary object problem in libpq++ from the
Bruce Momjian [Fri, 22 Oct 1999 19:05:02 +0000 (19:05 +0000)]
This *should* fix the large binary object problem in libpq++ from the
TODO list.

Vince

25 years agoI have a patch for postgresql-snapshot(1999-10-22).
Bruce Momjian [Fri, 22 Oct 1999 11:51:35 +0000 (11:51 +0000)]
I have a patch for postgresql-snapshot(1999-10-22).

This patch fix a TODO list item.
* require SELECT DISTINCT target list to have all ORDER BY columns

example
ogawa=> select distinct x from t1 order by y;
ERROR:  ORDER BY columns must appear in SELECT DISTINCT target list

---
Atsushi Ogawa

25 years agoUpdate TODO list.
Bruce Momjian [Fri, 22 Oct 1999 11:51:12 +0000 (11:51 +0000)]
Update TODO list.

25 years agoAnother day, another MAX_QUERY_SIZE dependency zapped.
Tom Lane [Fri, 22 Oct 1999 02:08:37 +0000 (02:08 +0000)]
Another day, another MAX_QUERY_SIZE dependency zapped.

25 years agoUpdate TODO list.
Bruce Momjian [Thu, 21 Oct 1999 16:58:49 +0000 (16:58 +0000)]
Update TODO list.

25 years agoBugfix for CREATE RULE invoked via SPI --- from Hiroshi,
Tom Lane [Thu, 21 Oct 1999 02:33:25 +0000 (02:33 +0000)]
Bugfix for CREATE RULE invoked via SPI --- from Hiroshi,
23 Jul 99.  Not sure why this didn't get applied before...

25 years agoRemove fixed-size buffers in rule storage routine.
Tom Lane [Thu, 21 Oct 1999 01:46:24 +0000 (01:46 +0000)]
Remove fixed-size buffers in rule storage routine.

25 years agochange issue to information in banner
Bruce Momjian [Thu, 21 Oct 1999 01:24:53 +0000 (01:24 +0000)]
change issue to information in banner

25 years agoMake psql startup banner cleaner.
Bruce Momjian [Thu, 21 Oct 1999 00:43:46 +0000 (00:43 +0000)]
Make psql startup banner cleaner.

25 years agoBufFileSeek's behavior at segment boundaries wasn't what
Tom Lane [Tue, 19 Oct 1999 02:34:45 +0000 (02:34 +0000)]
BufFileSeek's behavior at segment boundaries wasn't what
logfile.c wanted ... seems easier to fix BufFileSeek.

25 years agoUpdate \dT again.
Bruce Momjian [Mon, 18 Oct 1999 14:14:04 +0000 (14:14 +0000)]
Update \dT again.

25 years agoMore \dT cleanup
Bruce Momjian [Mon, 18 Oct 1999 14:07:49 +0000 (14:07 +0000)]
More \dT cleanup

25 years agoClean up \dT display.
Bruce Momjian [Mon, 18 Oct 1999 13:44:24 +0000 (13:44 +0000)]
Clean up \dT display.

25 years agoMake bpchar == char() clearer in \dT display.
Bruce Momjian [Mon, 18 Oct 1999 11:36:45 +0000 (11:36 +0000)]
Make bpchar == char() clearer in \dT display.

25 years agoDrive a stake through the heart of the last use of MAX_PARSE_BUFFER
Tom Lane [Mon, 18 Oct 1999 03:32:29 +0000 (03:32 +0000)]
Drive a stake through the heart of the last use of MAX_PARSE_BUFFER
in the backend.  Still a few stragglers, but we're getting closer to
being rid of query length limits...

25 years agoRemove fixed-size literal buffer from scan.l, and repair
Tom Lane [Mon, 18 Oct 1999 02:42:31 +0000 (02:42 +0000)]
Remove fixed-size literal buffer from scan.l, and repair
boundary-condition bug in myinput() which caused flex scanner to fail
on tokens larger than a bufferload.  Turns out flex doesn't want null-
terminated input ... and if it gives you a 1-character buffer, you'd
better supply a character, not a null, lest you be thought to be
reporting end of input.

25 years agoUpdate TODO list.
Bruce Momjian [Mon, 18 Oct 1999 01:12:51 +0000 (01:12 +0000)]
Update TODO list.

25 years agoExcept_Intersect_Rewrite() forgot to move LIMIT info to new
Tom Lane [Sun, 17 Oct 1999 23:50:43 +0000 (23:50 +0000)]
Except_Intersect_Rewrite() forgot to move LIMIT info to new
topmost SELECT node after rearranging query tree.

25 years agoChange fd.c so that temp files are closed and deleted at
Tom Lane [Sun, 17 Oct 1999 23:09:02 +0000 (23:09 +0000)]
Change fd.c so that temp files are closed and deleted at
proc_exit time.  I discovered that if the frontend closes the connection
when you're inside a transaction block, there is nothing ensuring that
temp files go away ... I wonder whether proc_exit ought to try to do an
explicit transaction abort?

25 years agoRemove now-dead sort modules.
Tom Lane [Sun, 17 Oct 1999 22:19:07 +0000 (22:19 +0000)]
Remove now-dead sort modules.

25 years agoFinal stage of psort reconstruction work: replace psort.c with
Tom Lane [Sun, 17 Oct 1999 22:15:09 +0000 (22:15 +0000)]
Final stage of psort reconstruction work: replace psort.c with
a generalized module 'tuplesort.c' that can sort either HeapTuples or
IndexTuples, and is not tied to execution of a Sort node.  Clean up
memory leakages in sorting, and replace nbtsort.c's private implementation
of mergesorting with calls to tuplesort.c.

25 years agoActually, nodeMergejoin shouldn't include psort.h at all...
Tom Lane [Sun, 17 Oct 1999 18:00:29 +0000 (18:00 +0000)]
Actually, nodeMergejoin shouldn't include psort.h at all...

25 years agonodeMergejoin was depending on an inclusion from psort.h...
Tom Lane [Sat, 16 Oct 1999 21:31:26 +0000 (21:31 +0000)]
nodeMergejoin was depending on an inclusion from psort.h...

25 years agoSecond phase of psort reconstruction project: add bookkeeping logic to
Tom Lane [Sat, 16 Oct 1999 19:49:28 +0000 (19:49 +0000)]
Second phase of psort reconstruction project: add bookkeeping logic to
recycle storage within sort temp file on a block-by-block basis.  This
reduces peak disk usage to essentially just the volume of data being
sorted, whereas it had been about 4x the data volume before.

25 years agoCheck RELSEG_SIZE when postmaster starting up.
Tatsuo Ishii [Sat, 16 Oct 1999 09:32:23 +0000 (09:32 +0000)]
Check RELSEG_SIZE when postmaster starting up.
this is neccesary to make sure that the backend and
the database uses same RELSEG_SIZE.

25 years ago*** empty log message ***
Michael Meskes [Fri, 15 Oct 1999 19:02:08 +0000 (19:02 +0000)]
*** empty log message ***

25 years agoFix typo in descriptions.
Bruce Momjian [Fri, 15 Oct 1999 16:19:40 +0000 (16:19 +0000)]
Fix typo in descriptions.

25 years agoClean up type descriptions.
Bruce Momjian [Fri, 15 Oct 1999 04:56:52 +0000 (04:56 +0000)]
Clean up type descriptions.

25 years agoThis patch implements ORACLE's COMMENT SQL command.
Bruce Momjian [Fri, 15 Oct 1999 01:49:49 +0000 (01:49 +0000)]
This patch implements ORACLE's COMMENT SQL command.
>From the ORACLE 7 SQL Language Reference Manual:
-----------------------------------------------------
COMMENT

Purpose:

To add a comment about a table, view, snapshot, or
column into the data dictionary.

Prerequisites:

The table, view, or snapshot must be in your own
schema
or you must have COMMENT ANY TABLE system privilege.

Syntax:

COMMENT ON [ TABLE table ] |
           [ COLUMN table.column] IS 'text'

You can effectively drop a comment from the database
by setting it to the empty string ''.
-----------------------------------------------------

Example:

COMMENT ON TABLE workorders IS
   'Maintains base records for workorder information';

COMMENT ON COLUMN workorders.hours IS
   'Number of hours the engineer worked on the task';

to drop a comment:

COMMENT ON COLUMN workorders.hours IS '';

The current patch will simply perform the insert into
pg_description, as per the TODO. And, of course, when
the table is dropped, any comments relating to it
or any of its attributes are also dropped. I haven't
looked at the ODBC source yet, but I do know from
an ODBC client standpoint that the standard does
support the notion of table and column comments.
Hopefully the ODBC driver is already fetching these
values from pg_description, but if not, it should be
trivial.

Hope this makes the grade,

Mike Mascari
(mascarim@yahoo.com)

25 years agoAdd S-tree doc for possible future use.
Bruce Momjian [Fri, 15 Oct 1999 01:47:55 +0000 (01:47 +0000)]
Add S-tree doc for possible future use.

25 years agoUpdate psql \? for \w.
Bruce Momjian [Thu, 14 Oct 1999 01:28:42 +0000 (01:28 +0000)]
Update psql \? for \w.

25 years agoCorrectly return -1 on error to CmdTuples()
Bruce Momjian [Wed, 13 Oct 1999 16:46:28 +0000 (16:46 +0000)]
Correctly return -1 on error to CmdTuples()

Vince.