OSDN Git Service

pg-rex/syncrep.git
24 years agoFix up tabbing for _int8 entry.
Thomas G. Lockhart [Tue, 15 Feb 2000 03:28:31 +0000 (03:28 +0000)]
Fix up tabbing for _int8 entry.

24 years agoAdd "DEC" as synonym for "DECIMAL".
Thomas G. Lockhart [Tue, 15 Feb 2000 03:26:38 +0000 (03:26 +0000)]
Add "DEC" as synonym for "DECIMAL".
Add "SESSION_USER" as SQL92 keyword; equivalent to CURRENT_USER for now.
Implement column aliases (aka correlation names) and more join syntax.
Fix up indenting and tabbing.

24 years agoRecognize special case of POSIX time zone: "GMT+8" and "GMT-8".
Thomas G. Lockhart [Tue, 15 Feb 2000 03:17:09 +0000 (03:17 +0000)]
Recognize special case of POSIX time zone: "GMT+8" and "GMT-8".
Still needs to be done for the general case:
 "tz+/-#" where tz is a 3 char string.
This will probably involve moving code around to other places.

24 years agoFix up error and log messages.
Thomas G. Lockhart [Tue, 15 Feb 2000 03:00:37 +0000 (03:00 +0000)]
Fix up error and log messages.

24 years agoI suspect that you are not the person to send this to, but I wasn't sure
Bruce Momjian [Mon, 14 Feb 2000 12:33:19 +0000 (12:33 +0000)]
I suspect that you are not the person to send this to, but I wasn't sure
where else to mail it. I am the maintainer of unixODBC, and we have a
set of code in our project that started life as the Postgres windows
ODBC driver, which has been ported back to unix. Anyway I have just
fixed a memory leak in the driver, and I cant see any mention of the fix
being done in the main Postgres code, so I thougth I would let you know.

Its in the statement.c module, after the COMMIT statement has been
executed in SC_Execute, the code was

Nick Gorham

24 years agoUpdate README.NT with new NT ipc location.
Bruce Momjian [Mon, 14 Feb 2000 12:29:58 +0000 (12:29 +0000)]
Update README.NT with new NT ipc location.

24 years ago*** empty log message ***
Michael Meskes [Mon, 14 Feb 2000 10:00:06 +0000 (10:00 +0000)]
*** empty log message ***

24 years agoAdjusted psql echoing options (-a and -e)
Peter Eisentraut [Sun, 13 Feb 2000 21:45:15 +0000 (21:45 +0000)]
Adjusted psql echoing options (-a and -e)

24 years agocontrib-array.patch
Bruce Momjian [Sun, 13 Feb 2000 18:59:53 +0000 (18:59 +0000)]
contrib-array.patch

        this is an old patch which I have already submitted and never seen
        in the sources. It corrects the datatype oids used in some iterator
        functions. This bug has been reported to me by many other people.

contrib-datetime.patch

        some code contributed by Reiner Dassing <dassing@wettzell.ifag.de>

contrib-makefiles.patch

        fixes all my contrib makefiles which don't work with some compilers,
        as reported to me by another user.

contrib-miscutil.patch

        an old patch for one of my old contribs.

contrib-string.patch

        a small change to the c-like text output functions. Now the '{'
        is escaped only at the beginning of the string to distinguish it
        from arrays, and the '}' is no more escaped.

elog-lineno.patch

        adds the current lineno of CopyFrom to elog messages. This is very
        useful when you load a 1 million tuples table from an external file
        and there is a bad value somehere. Currently you get an error message
        but you can't know where is the bad data. The patch uses a variable
        which was declared static in copy.c. The variable is now exported
        and initialized to 0. It is always cleared at the end of the copy
        or at the first elog message or when the copy is canceled.
        I know this is very ugly but I can't find any better way of knowing
        where the copy fails and I have this problem quite often.

plperl-makefile.patch

        fixes a typo in a makefile, but the error must be elsewhere because
        it is a file generated automatically. Please have a look.

tprintf-timestamp.patch

        restores the original 2-digit year format, assuming that the two
        century digits don't carry much information and that '000202' is
        easier to read than 20000202. Being only a log file it shouldn't
        break anything.

Please apply the patches before the next scheduled code freeze.

I also noticed that some of the contribs don't compile correcly. Should we
ask people to fix their code or rename their makefiles so that they are
ignored by the top makefile?

--
Massimo Dal Zotto

24 years ago2. trigger.c fails to compile due to a syntax error. It contains
Bruce Momjian [Sun, 13 Feb 2000 13:21:11 +0000 (13:21 +0000)]
2.  trigger.c fails to compile due to a syntax error.  It contains
    a switch statement that has an empty default label.  A label of a
    switch statement must be followed by a statement (or a label which
    is followed by a statement (or a label which ...)).

3.  Files include stringinfo.h failed to compile.  The macro,
    'appendStringInfoCharMacro' is implemented with a '?:' operation
    that returns a void expression for the true part and a char expresion
    for the false part.  Both the true and false parts of the '?:' oper-
    ator must return the same type.

Billy G. Allie

24 years agoDefault argument for createdb
Peter Eisentraut [Thu, 10 Feb 2000 20:08:58 +0000 (20:08 +0000)]
Default argument for createdb
Changed EXIT_ON_ERROR to ON_ERROR_STOP in psql

24 years agoAdd btree indexing of boolean values
Bruce Momjian [Thu, 10 Feb 2000 19:51:52 +0000 (19:51 +0000)]
Add btree indexing of boolean values
 Don Baccus

24 years agoImprove default compile ops.
Bruce Momjian [Thu, 10 Feb 2000 19:11:17 +0000 (19:11 +0000)]
Improve default compile ops.

24 years agoAdd no CVS to make_mkid.
Bruce Momjian [Thu, 10 Feb 2000 18:41:35 +0000 (18:41 +0000)]
Add no CVS to make_mkid.

24 years agoPrevent tags files in cvs directory.
Bruce Momjian [Thu, 10 Feb 2000 18:34:26 +0000 (18:34 +0000)]
Prevent tags files in cvs directory.

24 years agoYou will need to copy "config.h.win32" to "config.h" in the include
Bruce Momjian [Thu, 10 Feb 2000 15:47:58 +0000 (15:47 +0000)]
You will need to copy "config.h.win32" to "config.h" in the include
directory.

I think this patch to the docs should be what is needed.

 Magnus Hagander

24 years agoUpdate length to 31.
Bruce Momjian [Thu, 10 Feb 2000 04:08:25 +0000 (04:08 +0000)]
Update length to 31.

24 years agoHere is the patch to add plperl to droplang.
Bruce Momjian [Wed, 9 Feb 2000 20:23:26 +0000 (20:23 +0000)]
Here is the patch to add plperl to droplang.
Mark Hollomon

24 years agoUpdate TODO list.
Bruce Momjian [Wed, 9 Feb 2000 20:09:29 +0000 (20:09 +0000)]
Update TODO list.

24 years agoI checked the WinNT port yesterday (a few days old snapshot from CVS) and I
Bruce Momjian [Wed, 9 Feb 2000 16:23:44 +0000 (16:23 +0000)]
I checked the WinNT port yesterday (a few days old snapshot from CVS) and I
am including a patch to get it compile.

changes to psql:
- added less as default pager when compiling on Cygwin
- need to declare "filename_completion_function" because it is not exported
from readline -> added to include/port/win.h

changes to pg_id:
- include of <getopt.h>
- add .exe when installing

I think there is a problem with calling the regress tests on WinNT - it
should be called with PORTNAME not HOST as the parameter to regress.sh or
the check  when to add "-h localhost" to psql has to be changed. Now it is
checked against the PORTNAME.

The results of the regress tests were OK with expected failures ;-)

Daniel Horak

24 years agoEnable backward sequential scan even after reaching EOF.
Hiroshi Inoue [Wed, 9 Feb 2000 03:49:47 +0000 (03:49 +0000)]
Enable backward sequential scan even after reaching EOF.

24 years agoFixed broken initdb --show patch
Peter Eisentraut [Wed, 9 Feb 2000 00:21:49 +0000 (00:21 +0000)]
Fixed broken initdb --show patch

24 years agoOoops ... 'char c' can hold a char, but it can't hold EOF ...
Tom Lane [Wed, 9 Feb 2000 00:10:11 +0000 (00:10 +0000)]
Ooops ... 'char c' can hold a char, but it can't hold EOF ...

24 years ago yesterday I had a problem with initdb. I set bad environment values
Bruce Momjian [Tue, 8 Feb 2000 15:58:54 +0000 (15:58 +0000)]
 yesterday I had a problem with initdb. I set bad environment values
and initdb crashs (I set pglib path to PG 6.5.3 directory instead to
7.0 and initdb take this BKI old templates ... (initdb not check
BKI version and BKI files not has any version comments (TODO?))

 This patch add to the initdb --show option which show setting of all
initdb's values. It spare developers time if in setting is bug.

                                                Karel

----------------------------------------------------------------------
Karel Zak <zakkr@zf.jcu.cz>              http://home.zf.jcu.cz/~zakkr/

24 years ago I'm sending patch with new version of to_char numbers formatting.
Bruce Momjian [Tue, 8 Feb 2000 15:57:01 +0000 (15:57 +0000)]
 I'm sending patch with new version of to_char numbers formatting.

 The PostgreSQL's to_char() is very compatible with Oracle's to_char
 now. I hope that to_char's 3000 rows of source is without bugs, but
 will good if anyone test it, for me it works very well :-)

                                                        Karel

----------------------------------------------------------------------
Karel Zak <zakkr@zf.jcu.cz>              http://home.zf.jcu.cz/~zakkr/

24 years agoFixed psql double quoting of SQL ids
Peter Eisentraut [Mon, 7 Feb 2000 23:10:11 +0000 (23:10 +0000)]
Fixed psql double quoting of SQL ids
Fixed libpq printing functions

24 years agoMask removal of network_ops in 7.0.
Bruce Momjian [Mon, 7 Feb 2000 21:24:15 +0000 (21:24 +0000)]
Mask removal of network_ops in 7.0.

24 years agoEnabled MATCH <unspecified>
Jan Wieck [Mon, 7 Feb 2000 18:12:49 +0000 (18:12 +0000)]
Enabled MATCH <unspecified>

Jan

24 years agoAdded complete MATCH <unspecified> support contributed by Don Baccus.
Jan Wieck [Mon, 7 Feb 2000 17:50:38 +0000 (17:50 +0000)]
Added complete MATCH <unspecified> support contributed by Don Baccus.

Jan

24 years agoDisable/reenable triggers for data only dump.
Jan Wieck [Mon, 7 Feb 2000 16:30:58 +0000 (16:30 +0000)]
Disable/reenable triggers for data only dump.

Jan

24 years agoRepair planning bugs caused by my misguided removal of restrictinfo link
Tom Lane [Mon, 7 Feb 2000 04:41:04 +0000 (04:41 +0000)]
Repair planning bugs caused by my misguided removal of restrictinfo link
fields in JoinPaths --- turns out that we do need that after all :-(.
Also, rearrange planner so that only one RelOptInfo is created for a
particular set of joined base relations, no matter how many different
subsets of relations it can be created from.  This saves memory and
processing time compared to the old method of making a bunch of RelOptInfos
and then removing the duplicates.  Clean up the jointree iteration logic;
not sure if it's better, but I sure find it more readable and plausible
now, particularly for the case of 'bushy plans'.

24 years agochange copyright year 1999->2000
Tatsuo Ishii [Mon, 7 Feb 2000 04:31:10 +0000 (04:31 +0000)]
change copyright year 1999->2000
Minor fix

24 years agoAllow SELECT .. FOR UPDATE in PL/pgSQL.
Hiroshi Inoue [Mon, 7 Feb 2000 03:39:13 +0000 (03:39 +0000)]
Allow SELECT .. FOR UPDATE in PL/pgSQL.

24 years agoRemove ugly call ImmediateSharedRelationCacheInvalidate()
Hiroshi Inoue [Mon, 7 Feb 2000 02:38:18 +0000 (02:38 +0000)]
Remove ugly call ImmediateSharedRelationCacheInvalidate()
from md.c.

24 years agoRemoved special hack causing ON UPDATE NO ACTION trigger to
Jan Wieck [Sun, 6 Feb 2000 10:19:45 +0000 (10:19 +0000)]
Removed special hack causing ON UPDATE NO ACTION trigger to
be suppressed.

Jan

24 years agoSolaris has always had problems with 1947 in the
Bruce Momjian [Sun, 6 Feb 2000 05:09:42 +0000 (05:09 +0000)]
Solaris has always had problems with 1947 in the
regression tests so I prepared a set of expected
files to make things look OK.

There's also a file to account for minor variations
in the geopmetry output and a resultmap patch to
pull them all together.

With these changes PostgreSQL, from CVS, builds and
regression tests (runcheck) cleanly.

Keith Parks.

24 years agoRename same() to sameseti() to have a slightly less generic name. Move
Tom Lane [Sun, 6 Feb 2000 03:27:35 +0000 (03:27 +0000)]
Rename same() to sameseti() to have a slightly less generic name.  Move
nonoverlap_sets() and is_subset() to list.c, where they should have lived
to begin with, and rename to nonoverlap_setsi and is_subseti since they
only work on integer lists.

24 years agoFix a bug when specfying -D option.
Tatsuo Ishii [Sun, 6 Feb 2000 00:49:03 +0000 (00:49 +0000)]
Fix a bug when specfying -D option.
fix suggested by Keith Parks.

24 years agoReplace nth() calls in inner loops with traversal of the list via
Tom Lane [Sat, 5 Feb 2000 23:19:44 +0000 (23:19 +0000)]
Replace nth() calls in inner loops with traversal of the list via
lnext, to eliminate O(N^2) behavior with lots of indexquals.

24 years agoFurther cleanup for OR-of-AND WHERE-clauses. orindxpath can now handle
Tom Lane [Sat, 5 Feb 2000 18:26:09 +0000 (18:26 +0000)]
Further cleanup for OR-of-AND WHERE-clauses.  orindxpath can now handle
extracting from an AND subclause just those opclauses that are relevant
for a particular index.  For example, we can now consider using an index
on x to process WHERE (x = 1 AND y = 2) OR (x = 2 AND y = 4) OR ...

24 years agoUpdate TODO list.
Bruce Momjian [Sat, 5 Feb 2000 17:02:43 +0000 (17:02 +0000)]
Update TODO list.

24 years agoChange function name PQclientencoding to PQclientEncoding since
Tatsuo Ishii [Sat, 5 Feb 2000 12:33:22 +0000 (12:33 +0000)]
Change function name PQclientencoding to PQclientEncoding since
it seems more suitable for the naming convention in libpq.

New function PQsetClientEncoding added. It makes possible to change
the client encoding on the fly without setting PGCLIENTENCODING.

24 years agoChange function name PQclientencoding to PQclientEncoding since
Tatsuo Ishii [Sat, 5 Feb 2000 12:27:56 +0000 (12:27 +0000)]
Change function name PQclientencoding to PQclientEncoding since
it seems more suitable for the naming convention in libpq.

24 years agoEnabling automatic primary key detection for self-referencing
Jan Wieck [Sat, 5 Feb 2000 00:20:38 +0000 (00:20 +0000)]
Enabling automatic primary key detection for self-referencing
FOREIGN KEY constraint during CREATE TABLE. Tnx to Stephan.

Jan

24 years agoSmall bugfix for DROP TABLE if table is self-referenced by
Jan Wieck [Fri, 4 Feb 2000 23:45:04 +0000 (23:45 +0000)]
Small bugfix for DROP TABLE if table is self-referenced by
a FOREIGN KEY constraint.

Jan

24 years agoUpdate TODO list.
Bruce Momjian [Fri, 4 Feb 2000 22:41:28 +0000 (22:41 +0000)]
Update TODO list.

24 years agoAdded ALTER TABLE ... ADD CONSTRAINT (provided by Stephan Szabo).
Jan Wieck [Fri, 4 Feb 2000 18:49:34 +0000 (18:49 +0000)]
Added ALTER TABLE ... ADD CONSTRAINT (provided by Stephan Szabo).

Added constraint dumping capability to pg_dump (also from Stephan)

Fixed DROP TABLE -> RelationBuildTriggers: 2 record(s) not found for rel
error.

Fixed little error in gram.y I made the last days.

Jan

24 years agoMade prototypes for yyerror() consistent with Bruce's last changes
Jan Wieck [Fri, 4 Feb 2000 18:45:23 +0000 (18:45 +0000)]
Made prototypes for yyerror() consistent with Bruce's last changes

Jan

24 years agoUpdate TODO list.
Bruce Momjian [Fri, 4 Feb 2000 05:04:55 +0000 (05:04 +0000)]
Update TODO list.

24 years agoSquash longstanding memory leak: when catcache.c copied a tuple into
Tom Lane [Fri, 4 Feb 2000 03:16:03 +0000 (03:16 +0000)]
Squash longstanding memory leak: when catcache.c copied a tuple into
the cache context, it didn't bother to free the tuple that
CatalogIndexFetchTuple had allocated in the transaction context.
Do enough cache lookups in the same xact, and you start to notice...

24 years agoUpdate TODO list.
Bruce Momjian [Thu, 3 Feb 2000 12:22:33 +0000 (12:22 +0000)]
Update TODO list.

24 years agoMake EXPLAIN results for Append, Group, Agg, Unique nodes more plausible.
Tom Lane [Thu, 3 Feb 2000 06:12:19 +0000 (06:12 +0000)]
Make EXPLAIN results for Append, Group, Agg, Unique nodes more plausible.
Group and Unique use an arbitrary assumption that there will be about
10% as many groups as input tuples --- perhaps someday we can refine this.

24 years agoRepair relation refcount leakage caused by SELECT FOR UPDATE.
Tom Lane [Thu, 3 Feb 2000 00:02:58 +0000 (00:02 +0000)]
Repair relation refcount leakage caused by SELECT FOR UPDATE.

24 years agoInconsistency in CREATE CONSTRAINT TRIGGER with the
Jan Wieck [Wed, 2 Feb 2000 20:54:17 +0000 (20:54 +0000)]
Inconsistency in CREATE CONSTRAINT TRIGGER with the
actions performed by analyse.c when creating table constraints.

Jan

24 years agoAdd short chapter in developer's guide on formatting source code.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:25:04 +0000 (16:25 +0000)]
Add short chapter in developer's guide on formatting source code.

24 years agoAdd information on QNX port from Andreas Kardos.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:24:19 +0000 (16:24 +0000)]
Add information on QNX port from Andreas Kardos.

24 years agoRemove extraneous end tag. Code now compiles cleanly.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:23:37 +0000 (16:23 +0000)]
Remove extraneous end tag. Code now compiles cleanly.

24 years agoAdd info on debian package installation of sgml toolkits.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:22:45 +0000 (16:22 +0000)]
Add info on debian package installation of sgml toolkits.

24 years agoAdd detailed information on Australian time zones.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:21:06 +0000 (16:21 +0000)]
Add detailed information on Australian time zones.

24 years agoUpdate with current summary information on pgadmin.
Thomas G. Lockhart [Wed, 2 Feb 2000 16:19:54 +0000 (16:19 +0000)]
Update with current summary information on pgadmin.
 From pgadmin author on 99-12-24.

24 years ago*** empty log message ***
Peter Eisentraut [Wed, 2 Feb 2000 13:20:15 +0000 (13:20 +0000)]
*** empty log message ***

24 years agoFixed bug in NO ACTION trigger proc entries.
Jan Wieck [Mon, 31 Jan 2000 14:02:27 +0000 (14:02 +0000)]
Fixed bug in NO ACTION trigger proc entries.

Jan

24 years agoFix problems seen in parallel regress tests when SI buffer overruns (causing
Tom Lane [Mon, 31 Jan 2000 04:35:57 +0000 (04:35 +0000)]
Fix problems seen in parallel regress tests when SI buffer overruns (causing
syscache and relcache flushes).  Relcache entry rebuild now preserves
original tupledesc, rewrite rules, and triggers if possible, so that pointers
to these things remain valid --- if these things change while relcache entry
has positive refcount, we elog(ERROR) to avoid later crash.  Arrange for
xact-local rels to be rebuilt when an SI inval message is seen for them,
so that they are updated by CommandCounterIncrement the same as regular rels.
(This is useful because of Hiroshi's recent changes to process our own SI
messages at CommandCounterIncrement time.)  This allows simplification of
some routines that previously hacked around the lack of an automatic update.
catcache now keeps its own copy of tupledesc for its relation, rather than
depending on the relcache's copy; this avoids needing to reinitialize catcache
during a cache flush, which saves some cycles and eliminates nasty circularity
problems that occur if a cache flush happens while trying to initialize a
catcache.
Eliminate a number of permanent memory leaks that used to happen during
catcache or relcache flush; not least of which was that catcache never
freed any cached tuples!  (Rule parsetree storage is still leaked, however;
will fix that separately.)
Nothing done yet about code that uses tuples retrieved by SearchSysCache
for longer than is safe.

24 years agoHmm, equalfuncs didn't know about SortClause or GroupClause
Tom Lane [Mon, 31 Jan 2000 01:21:39 +0000 (01:21 +0000)]
Hmm, equalfuncs didn't know about SortClause or GroupClause
nodes...

24 years agoAdded numeric and int8 types.
Byron Nikolaidis [Sun, 30 Jan 2000 21:14:25 +0000 (21:14 +0000)]
Added numeric and int8 types.
Bug fix for LongVarBinary -- begin transaction

24 years agoDelete unused and long-dead header file.
Tom Lane [Sat, 29 Jan 2000 21:13:23 +0000 (21:13 +0000)]
Delete unused and long-dead header file.

24 years agoModify uses of RelationFlushRelation and RelationCacheInvalidate so that
Tom Lane [Sat, 29 Jan 2000 19:51:59 +0000 (19:51 +0000)]
Modify uses of RelationFlushRelation and RelationCacheInvalidate so that
we *always* rebuild, rather than deleting, an invalidated relcache entry
that has positive refcount.  Otherwise an SI cache overrun leads to
dangling Relation pointers all over the place!

24 years agoUpdate TODO list.
Bruce Momjian [Sat, 29 Jan 2000 17:25:46 +0000 (17:25 +0000)]
Update TODO list.

24 years agoA few minor psql enhancements
Peter Eisentraut [Sat, 29 Jan 2000 16:58:54 +0000 (16:58 +0000)]
A few minor psql enhancements
Initdb help correction
Changed end/abort to commit/rollback and changed related notices
Commented out way old printing functions in libpq
Fixed a typo in alter table / alter column

24 years agoUpdate copyright file.
Bruce Momjian [Sat, 29 Jan 2000 08:53:10 +0000 (08:53 +0000)]
Update copyright file.

24 years agoAttached is a uuencoded tarball that contains
Bruce Momjian [Sat, 29 Jan 2000 01:58:50 +0000 (01:58 +0000)]
Attached is a uuencoded tarball that contains
3 new files and two patches for the plperl subdir.

These changes add the ability for plperl functions
to call 'elog'. It also sets up the frame work to
allow me to add access to the SPI functions.

--
Mark Hollomon

24 years agoPatch from Hiroshi for overflow btree comparison.
Bruce Momjian [Fri, 28 Jan 2000 17:23:47 +0000 (17:23 +0000)]
Patch from Hiroshi for overflow btree comparison.

24 years agoUpdate TODO list.
Bruce Momjian [Fri, 28 Jan 2000 04:47:49 +0000 (04:47 +0000)]
Update TODO list.

24 years agoUpdate TODO list.
Bruce Momjian [Fri, 28 Jan 2000 03:46:06 +0000 (03:46 +0000)]
Update TODO list.

24 years agoUpdate TODO list.
Bruce Momjian [Fri, 28 Jan 2000 03:40:34 +0000 (03:40 +0000)]
Update TODO list.

24 years agoApply the heuristic proposed by Taral (see pgsql-general archives for
Tom Lane [Fri, 28 Jan 2000 03:22:36 +0000 (03:22 +0000)]
Apply the heuristic proposed by Taral (see pgsql-general archives for
2-Oct-98 or TODO.detail/cnfify) to decide whether we want to reduce
WHERE clause to CNF form, DNF form, or neither.  This is a HUGE win.
The heuristic conditions could probably still use a little tweaking to
make sure we don't pick CNF when DNF would be better, or vice versa,
but the risk of exponential explosion in cnfify() is gone.  I was able
to run ten-thousand-AND-subclause queries through the planner in a
reasonable amount of time.

24 years ago*** empty log message ***
Michael Meskes [Thu, 27 Jan 2000 19:01:35 +0000 (19:01 +0000)]
*** empty log message ***

24 years agoRedesign DISTINCT ON as discussed in pgsql-sql 1/25/00: syntax is now
Tom Lane [Thu, 27 Jan 2000 18:11:50 +0000 (18:11 +0000)]
Redesign DISTINCT ON as discussed in pgsql-sql 1/25/00: syntax is now
SELECT DISTINCT ON (expr [, expr ...]) targetlist ...
and there is a check to make sure that the user didn't specify an ORDER BY
that's incompatible with the DISTINCT operation.
Reimplement nodeUnique and nodeGroup to use the proper datatype-specific
equality function for each column being compared --- they used to do
bitwise comparisons or convert the data to text strings and strcmp().
(To add insult to injury, they'd look up the conversion functions once
for each tuple...)  Parse/plan representation of DISTINCT is now a list
of SortClause nodes.
initdb forced by querytree change...

24 years agofix for compile.
Bruce Momjian [Thu, 27 Jan 2000 05:33:51 +0000 (05:33 +0000)]
fix for compile.

24 years agoRemove duplicate extern declaration.
Tom Lane [Wed, 26 Jan 2000 23:48:05 +0000 (23:48 +0000)]
Remove duplicate extern declaration.

24 years agoUpdate files.
Bruce Momjian [Wed, 26 Jan 2000 16:10:01 +0000 (16:10 +0000)]
Update files.

24 years agoAdd include for float.h.
Bruce Momjian [Wed, 26 Jan 2000 06:33:49 +0000 (06:33 +0000)]
Add include for float.h.

24 years agoAdd:
Bruce Momjian [Wed, 26 Jan 2000 05:58:53 +0000 (05:58 +0000)]
Add:

  * Portions Copyright (c) 1996-2000, PostgreSQL, Inc

to all files copyright Regents of Berkeley.  Man, that's a lot of files.

24 years ago as attache of this mail is patch (to the main tree) with to_char's
Bruce Momjian [Tue, 25 Jan 2000 23:53:56 +0000 (23:53 +0000)]
 as attache of this mail is patch (to the main tree) with to_char's
family functions. Contain:

  conversion from a datetype to formatted text:

to_char( datetime,  text)
to_char( timestamp, text)
to_char( int4, text)
to_char( int8, text)
to_char( float4, text)
to_char( float8, text)
to_char( numeric, text)

  vice versa:

to_date ( text, text)
to_datetime ( text, text)
to_timestamp ( text, text)
to_number ( text, text)    (convert to numeric)

  PostgreSQL to_char is very compatible with Oracle's to_char(), but not
total exactly (now). Small differentions are in number formating. It will
fix in next to_char() version.

! If will this patch aplly to the main tree, must be delete the current
  to_char version in contrib (directory "dateformat" and note in contrib's
  README), this patch not erase it (sorry Bruce).

The patch patching files:

doc/src/sgml/func.sgml
                     ^^^^^^^^
   Hmm, I'm not sure if my English... :( Check it anyone (volunteer)?

   Thomas, it is right? SGML is not my primary lang  and compile
   the current PG docs tree is very happy job (hard variables setting in
   docs/sgml/Makefile --> HSTYLE= /home/users/t/thomas/....  :-)

   What add any definition to global configure.in and set Makefiles in docs
   tree via ./configure?

src/backend/utils/adt/Makefile
src/backend/utils/adt/formatting.c
src/include/catalog/pg_proc.h
src/include/utils/formatting.h
Karel Zak <zakkr@zf.jcu.cz>              http://home.zf.jcu.cz/~zakkr/

24 years agoRemove obsolete references to 8K query limit.
Tom Lane [Tue, 25 Jan 2000 07:04:04 +0000 (07:04 +0000)]
Remove obsolete references to 8K query limit.

24 years agoFix pg_euccn_mblen() so that it always returns 2 if data is not ascii.
Tatsuo Ishii [Tue, 25 Jan 2000 02:12:27 +0000 (02:12 +0000)]
Fix pg_euccn_mblen() so that it always returns 2 if data is not ascii.
(EUC_CN does have only code set 0 and 1)

24 years agoForgot that expr assumes regexp is anchored-left; therefore need .*
Tom Lane [Tue, 25 Jan 2000 01:59:27 +0000 (01:59 +0000)]
Forgot that expr assumes regexp is anchored-left; therefore need .*
in front of any platform name pattern that's not supposed to match
beginning at the start of the machine type name...

24 years agoRemoved alter table drop column.
Peter Eisentraut [Mon, 24 Jan 2000 23:40:35 +0000 (23:40 +0000)]
Removed alter table drop column.

24 years agoUpdate TODO list.
Bruce Momjian [Mon, 24 Jan 2000 22:27:59 +0000 (22:27 +0000)]
Update TODO list.

24 years agoMade abstime/reltime use int4 instead of time_t (TODO item)
Peter Eisentraut [Mon, 24 Jan 2000 19:34:19 +0000 (19:34 +0000)]
Made abstime/reltime use int4 instead of time_t (TODO item)
Made type equivalency apply to aggregates (TODO item)
Fixed parsing bug in psql
Reverted some stupid options changes I made to pg_dump

24 years agoFirst cut at making useful selectivity estimates for range queries
Tom Lane [Mon, 24 Jan 2000 07:16:52 +0000 (07:16 +0000)]
First cut at making useful selectivity estimates for range queries
(ie, WHERE x > lowbound AND x < highbound).  It's not very bright yet
but it does something useful.  Also, rename intltsel/intgtsel to
scalarltsel/scalargtsel to reflect usage better.  Extend convert_to_scalar
to do something a little bit useful with string data types.  Still need
to make it do something with date/time datatypes, but I'll wait for
Thomas's datetime unification dust to settle first.  Eventually the
routine ought not have any type-specific knowledge at all; it ought to
be calling a type-dependent routine found via a pg_type column; but
that's a task for another day.

24 years agoUpdate for index change. Semes it didn't work the first time.
Bruce Momjian [Mon, 24 Jan 2000 03:08:27 +0000 (03:08 +0000)]
Update for index change.  Semes it didn't work the first time.

24 years agoRemove Ops parameter from STATRELID cache lookup, for Tom Lane and
Bruce Momjian [Mon, 24 Jan 2000 02:12:58 +0000 (02:12 +0000)]
Remove Ops parameter from STATRELID cache lookup, for Tom Lane and
optimizer.

24 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Jan 2000 17:30:18 +0000 (17:30 +0000)]
Update TODO list.

24 years agoChange format of postmaster.opts.default.
Tatsuo Ishii [Sun, 23 Jan 2000 08:20:24 +0000 (08:20 +0000)]
Change format of postmaster.opts.default.
Executable file name (postmaster) no more included in the file.

24 years agoFix error messages for the lack of multi-byte support.
Tatsuo Ishii [Sun, 23 Jan 2000 08:16:37 +0000 (08:16 +0000)]
Fix error messages for the lack of multi-byte support.
Since --with-mb has been removed from configure, previous
messages were not appropriate.

24 years agochar_length()/octet_length for char() type now returns length of
Tatsuo Ishii [Sun, 23 Jan 2000 08:13:34 +0000 (08:13 +0000)]
char_length()/octet_length for char() type now returns length of
the charcter including trailing blanks.

24 years agoReplace SearchSysCacheGetAttribute with SysCacheGetAttr, which fetches
Tom Lane [Sun, 23 Jan 2000 03:43:24 +0000 (03:43 +0000)]
Replace SearchSysCacheGetAttribute with SysCacheGetAttr, which fetches
an attribute of a tuple previously fetched with SearchSysCacheTuple.
This avoids a lot of redundant cache lookups, particularly in selfuncs.c.
Also, remove SearchSysCacheStruct, which was unused and grotty.

24 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Jan 2000 03:28:54 +0000 (03:28 +0000)]
Update TODO list.

24 years agoUpdate TODO list.
Bruce Momjian [Sun, 23 Jan 2000 03:19:05 +0000 (03:19 +0000)]
Update TODO list.