]> granicus.if.org Git - postgresql/log
postgresql
28 years agoMake sure the btree patch gets into 2.0 as well...
Marc G. Fournier [Fri, 25 Oct 1996 09:55:36 +0000 (09:55 +0000)]
Make sure the btree patch gets into 2.0 as well...

Still submitted by:  Massimo Dal Zotto <dz@cs.unitn.it>

28 years agoAdd CFLAGS+=-DPOSTPORT=... so we can take it out of the global CFLAGS.
Bryan Henderson [Fri, 25 Oct 1996 09:22:30 +0000 (09:22 +0000)]
Add CFLAGS+=-DPOSTPORT=... so we can take it out of the global CFLAGS.

28 years agoUse EUROPEAN_DATES instead of EUROPEAN_STYLE
Bryan Henderson [Fri, 25 Oct 1996 06:02:52 +0000 (06:02 +0000)]
Use EUROPEAN_DATES instead of EUROPEAN_STYLE

28 years agoE Use EUROPEAN_DATES from config.h instead of -D option
Bryan Henderson [Fri, 25 Oct 1996 06:00:49 +0000 (06:00 +0000)]
E Use EUROPEAN_DATES from config.h instead of -D option

28 years agoFixes: Growing backend when using nested function calls
Marc G. Fournier [Thu, 24 Oct 1996 07:59:46 +0000 (07:59 +0000)]
Fixes: Growing backend when using nested function calls
Submitted by: wieck@sapserv.debis.de (Jan Wieck)

28 years agoFixes: Errors when PQexec() in backend creates temp
Marc G. Fournier [Thu, 24 Oct 1996 07:55:54 +0000 (07:55 +0000)]
Fixes: Errors when PQexec() in backend creates temp
              relations and transaction is aborted

Submitted by: wieck@sapserv.debis.de (Jan Wieck)

28 years agoFixes:
Marc G. Fournier [Thu, 24 Oct 1996 06:32:01 +0000 (06:32 +0000)]
Fixes:

It's bug in nodeAgg.c on lines 241, 242:

                null_array = malloc(nagg);
                for (i=0;i<nagg;i++)
                    null_array[i] = 'n';
                oneTuple = heap_formtuple(tupType, tupValue, null_array);

- your query has not only aggregates but also 'group by-ed' fields and so
null_array should contain tupType->natts elements (tupType->natts > nagg in
your case).

Patch follows and it's very simple.

VAdim

28 years agoTake out the PERFECT_MMGR #ifdefs:
Marc G. Fournier [Thu, 24 Oct 1996 06:30:40 +0000 (06:30 +0000)]
Take out the PERFECT_MMGR #ifdefs:

My guess is that the thing had bugs, and the pfree was commented out.
The thing is probabally free'ed anyway at the end, so it was not a bad
thing.

If it does cause a bug, it will generate an error when hit, so I say
unless someone else knows, let's remove it and run the regression test.

-Bruce

28 years agoBruce...does this fix it once and for all? *grin*
Marc G. Fournier [Thu, 24 Oct 1996 06:25:34 +0000 (06:25 +0000)]
Bruce...does this fix it once and for all? *grin*

28 years agoMajor code cleanups from D'arcy (-Wall -Werror)
Marc G. Fournier [Wed, 23 Oct 1996 07:42:13 +0000 (07:42 +0000)]
Major code cleanups from D'arcy (-Wall -Werror)

28 years ago|I have moved things around so that things happen in the following order:
Marc G. Fournier [Wed, 23 Oct 1996 07:33:04 +0000 (07:33 +0000)]
|I have moved things around so that things happen in the following order:
|
|  1. Set default variables
|  2. Include Makefile.custom to override defaults
|  3. Set CFLAGS, etc. with variables
|
|This fixes the problem of Makefile.custom changes not taking effect.

Submitted by: D'Arcy Cain

28 years agoUpdate "planner" directory to "optimizer" in include paths in this dead code.
Bryan Henderson [Wed, 23 Oct 1996 07:14:43 +0000 (07:14 +0000)]
Update "planner" directory to "optimizer" in include paths in this dead code.

28 years agoCosmetic changes to ordering of #include files
Marc G. Fournier [Mon, 21 Oct 1996 11:49:41 +0000 (11:49 +0000)]
Cosmetic changes to ordering of #include files

28 years ago-Wall'd
Marc G. Fournier [Mon, 21 Oct 1996 09:37:26 +0000 (09:37 +0000)]
-Wall'd

28 years ago-Wall'd
Marc G. Fournier [Mon, 21 Oct 1996 08:31:23 +0000 (08:31 +0000)]
-Wall'd

28 years ago-Wall'd ... I missed a subdir under access :(
Marc G. Fournier [Mon, 21 Oct 1996 07:38:20 +0000 (07:38 +0000)]
-Wall'd ... I missed a subdir under access :(

28 years agoMake line 159:bp = NULL, to quiet compiler
Marc G. Fournier [Mon, 21 Oct 1996 07:18:08 +0000 (07:18 +0000)]
Make line 159:bp = NULL, to quiet compiler

Suggested by: Randy Terbush <randy@zyzzyva.com>

28 years ago-Wall'd
Marc G. Fournier [Mon, 21 Oct 1996 07:15:18 +0000 (07:15 +0000)]
-Wall'd

That finishes access/*

28 years agoForce user to set PORTNAME. Remove NAMEDATALEN from CFLAGS (don't need it
Bryan Henderson [Mon, 21 Oct 1996 06:56:57 +0000 (06:56 +0000)]
Force user to set PORTNAME.  Remove NAMEDATALEN from CFLAGS (don't need it
there because it's in config.h, which is cleaner).

28 years ago-Wall'd
Marc G. Fournier [Mon, 21 Oct 1996 05:59:49 +0000 (05:59 +0000)]
-Wall'd

28 years ago-Wall'd
Marc G. Fournier [Mon, 21 Oct 1996 05:45:21 +0000 (05:45 +0000)]
-Wall'd

28 years ago-Wall Cleanup of gist subdirectory
Marc G. Fournier [Mon, 21 Oct 1996 05:14:02 +0000 (05:14 +0000)]
-Wall Cleanup of gist subdirectory

28 years ago-Wall cleanup of gist subdirectory:
Marc G. Fournier [Mon, 21 Oct 1996 05:11:00 +0000 (05:11 +0000)]
-Wall cleanup of gist subdirectory:
- cleaned out unused variables
- added missing prototype headers

One outstanding -Werror failure:

/home/staff/scrappy/postgres/2.0/cvs/postgres95/src/backend/access/gist/giststra
t.c:117: warning: missing braces around initializer for `GISTEvaluationData.expr
ession'

28 years agoDefine prototypes from giststrat.c required in gistscan.c
Marc G. Fournier [Mon, 21 Oct 1996 05:05:42 +0000 (05:05 +0000)]
Define prototypes from giststrat.c required in gistscan.c

28 years agoOther then:
Marc G. Fournier [Sun, 20 Oct 1996 22:04:49 +0000 (22:04 +0000)]
Other then:

indextuple.c:159: warning: `bp' might be used uninitialized in this function

this directory passes -Wall -Werror under FreeBSD

28 years agoremove an #ifdef/#else/#endif that doesn't actually have anything inside
Marc G. Fournier [Sun, 20 Oct 1996 21:51:45 +0000 (21:51 +0000)]
remove an #ifdef/#else/#endif that doesn't actually have anything inside
of it

28 years agoMake default HAVE_MEMMOVE defined, undef inside of definitions for sparc
Marc G. Fournier [Sun, 20 Oct 1996 20:58:14 +0000 (20:58 +0000)]
Make default HAVE_MEMMOVE defined, undef inside of definitions for sparc

28 years agochange '#if defined(PORTNAME_sparc)' to '#if !defined(HAVE_MEMMOVE)'
Marc G. Fournier [Sun, 20 Oct 1996 20:57:06 +0000 (20:57 +0000)]
change '#if defined(PORTNAME_sparc)' to '#if !defined(HAVE_MEMMOVE)'

28 years agoAdd #define HAVE_MEMMOVE for systems that do have it.
Marc G. Fournier [Sun, 20 Oct 1996 20:40:17 +0000 (20:40 +0000)]
Add #define HAVE_MEMMOVE for systems that do have it.

28 years agoClean up comments (-Wall -Werror)
Marc G. Fournier [Sun, 20 Oct 1996 20:34:06 +0000 (20:34 +0000)]
Clean up comments (-Wall -Werror)

28 years ago#include file cleanup
Marc G. Fournier [Sun, 20 Oct 1996 10:53:18 +0000 (10:53 +0000)]
#include file cleanup

28 years agoMore #include cleanups
Marc G. Fournier [Sun, 20 Oct 1996 09:27:24 +0000 (09:27 +0000)]
More #include cleanups

Once access/* is cleaned out, will redo using -Wall on compile to
make sure that all prototyping is correct

28 years agoFirst pass at fixing my own mistakes
Marc G. Fournier [Sun, 20 Oct 1996 08:32:11 +0000 (08:32 +0000)]
First pass at fixing my own mistakes

Mainly...fix up the includes I removed, as well as prototypes

Pointed out by D'Arcy

28 years agoAnother directory cleaned out
Marc G. Fournier [Sun, 20 Oct 1996 06:56:04 +0000 (06:56 +0000)]
Another directory cleaned out

28 years agoMore include files cleaned up
Marc G. Fournier [Sun, 20 Oct 1996 06:35:27 +0000 (06:35 +0000)]
More include files cleaned up

28 years agoOkay...another directory cleaned out and compilable
Marc G. Fournier [Sun, 20 Oct 1996 06:34:30 +0000 (06:34 +0000)]
Okay...another directory cleaned out and compilable

28 years agoAdd HAVE_Cplusplus= true to BSD44_derived Makefile.inc, as I know that
Marc G. Fournier [Sat, 19 Oct 1996 06:36:52 +0000 (06:36 +0000)]
Add HAVE_Cplusplus= true to BSD44_derived Makefile.inc, as I know that
FreeBSD/NetBSD both ship with gcc 2.7.x

28 years agoNot quite the cleanest way to do it, but add an option to Makefile.global
Marc G. Fournier [Sat, 19 Oct 1996 06:33:55 +0000 (06:33 +0000)]
Not quite the cleanest way to do it, but add an option to Makefile.global
that allows one to compile libpq++

this should be set on a port-by-port basis

28 years agoWelp, another subdirectory cleaned out of redundant/unused #include
Marc G. Fournier [Sat, 19 Oct 1996 06:27:33 +0000 (06:27 +0000)]
Welp, another subdirectory cleaned out of redundant/unused #include
files

28 years agoThere...that pretty much cleans up redundant/unused #includes in
Marc G. Fournier [Sat, 19 Oct 1996 04:51:44 +0000 (04:51 +0000)]
There...that pretty much cleans up redundant/unused #includes in
access/common...how many more directories to go? :)

28 years agoAlmost have access/common cleaned up...
Marc G. Fournier [Sat, 19 Oct 1996 04:49:29 +0000 (04:49 +0000)]
Almost have access/common cleaned up...

28 years agoA few more yet...
Marc G. Fournier [Sat, 19 Oct 1996 04:48:30 +0000 (04:48 +0000)]
A few more yet...

28 years agowish there was an easier way to clean this up...
Marc G. Fournier [Sat, 19 Oct 1996 04:46:28 +0000 (04:46 +0000)]
wish there was an easier way to clean this up...

28 years agoanother small one
Marc G. Fournier [Sat, 19 Oct 1996 04:36:08 +0000 (04:36 +0000)]
another small one

28 years agoHopefully that cleans up the cruft from rel.h
Marc G. Fournier [Sat, 19 Oct 1996 04:31:45 +0000 (04:31 +0000)]
Hopefully that cleans up the cruft from rel.h

28 years agoanother...
Marc G. Fournier [Sat, 19 Oct 1996 04:25:53 +0000 (04:25 +0000)]
another...

28 years agoits coming...
Marc G. Fournier [Sat, 19 Oct 1996 04:24:57 +0000 (04:24 +0000)]
its coming...

28 years agostep by step...file by file...
Marc G. Fournier [Sat, 19 Oct 1996 04:16:04 +0000 (04:16 +0000)]
step by step...file by file...

28 years ago...same
Marc G. Fournier [Sat, 19 Oct 1996 04:09:32 +0000 (04:09 +0000)]
...same

28 years agoslow work...
Marc G. Fournier [Sat, 19 Oct 1996 04:08:58 +0000 (04:08 +0000)]
slow work...

28 years agoand another
Marc G. Fournier [Sat, 19 Oct 1996 04:06:48 +0000 (04:06 +0000)]
and another

28 years agoanother one cleaned up
Marc G. Fournier [Sat, 19 Oct 1996 04:05:44 +0000 (04:05 +0000)]
another one cleaned up

28 years agoand another one..
Marc G. Fournier [Sat, 19 Oct 1996 04:03:29 +0000 (04:03 +0000)]
and another one..

28 years agoanother another one
Marc G. Fournier [Sat, 19 Oct 1996 04:02:39 +0000 (04:02 +0000)]
another another one

28 years agoanothre one reduced to its component parts
Marc G. Fournier [Sat, 19 Oct 1996 04:01:41 +0000 (04:01 +0000)]
anothre one reduced to its component parts

28 years agoclean out include files of including other includes files except those
Marc G. Fournier [Sat, 19 Oct 1996 03:58:34 +0000 (03:58 +0000)]
clean out include files of including other includes files except those
absolutely required

28 years agoAnother cleaned up
Marc G. Fournier [Sat, 19 Oct 1996 03:39:04 +0000 (03:39 +0000)]
Another cleaned up

28 years agoOne at a time...another one is complete...
Marc G. Fournier [Sat, 19 Oct 1996 03:28:20 +0000 (03:28 +0000)]
One at a time...another one is complete...

28 years agoanother one with #include file clean'd up
Marc G. Fournier [Sat, 19 Oct 1996 03:12:50 +0000 (03:12 +0000)]
another one with #include file  clean'd up

28 years agoAll #include's removed from *.h files, so cleaning up the .c #includes...
Marc G. Fournier [Fri, 18 Oct 1996 19:01:16 +0000 (19:01 +0000)]
All #include's removed from *.h files, so cleaning up the .c #includes...

First file of, what...1000's?

28 years ago#include "postgres.h" is a standard include for all .c files, auto-including
Marc G. Fournier [Fri, 18 Oct 1996 18:28:33 +0000 (18:28 +0000)]
#include "postgres.h" is a standard include for all .c files, auto-including
config.h, c.h *and* elog.h...added elog.h

28 years agoAdd prototype for heap_getattr() to quiet compiler
Marc G. Fournier [Fri, 18 Oct 1996 17:58:33 +0000 (17:58 +0000)]
Add prototype for heap_getattr() to quiet compiler

28 years agoReap deleted/unused tuples on page-base
Vadim B. Mikheev [Fri, 18 Oct 1996 08:15:58 +0000 (08:15 +0000)]
Reap deleted/unused tuples on page-base

28 years ago1. Reap deleted/unused tuples on page-base
Vadim B. Mikheev [Fri, 18 Oct 1996 08:13:36 +0000 (08:13 +0000)]
1. Reap deleted/unused tuples on page-base
2. Reap unused tuples too
3. Reap empty pages
4. Check if a page is initialized, initialize it if not
   and reap it
5. Binary search in list of reapped pages/tids to check
   is the heap' tid pointed by a index' tuple on this list
   (it's mu-u-uch faster)

28 years agoMove #define PageIsNew (true if page is not initialize) to bufpage.h
Vadim B. Mikheev [Fri, 18 Oct 1996 07:43:43 +0000 (07:43 +0000)]
Move #define PageIsNew (true if page is not initialize) to bufpage.h

28 years agoMove definition of PageIsNew from hio.c to here
Vadim B. Mikheev [Fri, 18 Oct 1996 07:39:07 +0000 (07:39 +0000)]
Move definition of PageIsNew from hio.c to here

28 years agoDocument some more #ifdef's
Marc G. Fournier [Fri, 18 Oct 1996 05:59:52 +0000 (05:59 +0000)]
Document some more #ifdef's

28 years agoDocument more #ifdef's into config.h
Marc G. Fournier [Fri, 18 Oct 1996 05:59:17 +0000 (05:59 +0000)]
Document more #ifdef's into config.h

Get rid of ESCAPE_PATCH ifdef, as its on by default, and there is no
apparent reason for turning it off...it fixes a bug

28 years agoMoved definition of MAXPATHLEN to config.h
Marc G. Fournier [Fri, 18 Oct 1996 05:50:04 +0000 (05:50 +0000)]
Moved definition of MAXPATHLEN to config.h

28 years agoAdded a #define NEED_SIG_JMP to config.h
Marc G. Fournier [Fri, 18 Oct 1996 05:47:12 +0000 (05:47 +0000)]
Added a #define NEED_SIG_JMP to config.h

Removed #ifdef's for OPENLINK_PATCH.  We enable it by default, and nobody
*has* to enable the functionality, but no reason to retain the "old code"

28 years agoDocument #ifdef's from backend/access subdirs so that we know what
Marc G. Fournier [Fri, 18 Oct 1996 05:22:24 +0000 (05:22 +0000)]
Document #ifdef's from backend/access subdirs so that we know what
features(?) are available

28 years agoMoved '#define *DEBUG' defines to config.h, and document where first
Marc G. Fournier [Fri, 18 Oct 1996 05:21:27 +0000 (05:21 +0000)]
Moved '#define *DEBUG' defines to config.h, and document where first
found.

Document any '#ifdef' segments found in config.h

28 years agoMoved '#define *DEBUG' defines to config.h, and document where first
Marc G. Fournier [Fri, 18 Oct 1996 05:21:09 +0000 (05:21 +0000)]
Moved '#define *DEBUG' defines to config.h, and document where first
found.

Document any '#ifdef' segments found in config.h

Remove gistold.c, since it was older copy of gist.c

28 years ago#define cleanup: moved #define IPORTAL_DEBUG to config.h
Marc G. Fournier [Fri, 18 Oct 1996 04:54:47 +0000 (04:54 +0000)]
#define cleanup: moved #define IPORTAL_DEBUG to config.h

28 years agoMoved all the CFLAG -D's from the makefiles to config.h towards better
Marc G. Fournier [Fri, 18 Oct 1996 04:49:09 +0000 (04:49 +0000)]
Moved all the CFLAG -D's from the makefiles to config.h towards better
centralized configurations/porting...

28 years agoMoved more -D's out of the port specific Makefiles straight into the
Marc G. Fournier [Fri, 18 Oct 1996 04:45:54 +0000 (04:45 +0000)]
Moved more -D's out of the port specific Makefiles straight into the
config.h file

28 years agoCentralize more of the defines from backend/port under config.h
Marc G. Fournier [Fri, 18 Oct 1996 00:33:41 +0000 (00:33 +0000)]
Centralize more of the defines from backend/port under config.h

Further ports should *really* avoid backend/port and make use of
config.h

28 years agoMove "port specific" #ifdefs out of here and into config.h
Marc G. Fournier [Thu, 17 Oct 1996 23:59:45 +0000 (23:59 +0000)]
Move "port specific" #ifdefs out of here and into config.h
- created a HAVE_TZSET define for this purpose

28 years agoAdd a HAVE_TZSET define to clean up defines in backend/utils/adt/nabstime.c
Marc G. Fournier [Thu, 17 Oct 1996 23:55:27 +0000 (23:55 +0000)]
Add a HAVE_TZSET define to clean up defines in backend/utils/adt/nabstime.c

28 years agoChange treatment of sigaction structure again to include even more platforms.
Bryan Henderson [Wed, 16 Oct 1996 09:41:13 +0000 (09:41 +0000)]
Change treatment of sigaction structure again to include even more platforms.
Thanks D'Arcy.

28 years agoRemove 4th element of sigaction structure so it works on more platforms.
Bryan Henderson [Tue, 15 Oct 1996 07:16:41 +0000 (07:16 +0000)]
Remove 4th element of sigaction structure so it works on more platforms.

28 years agoNew FAQ's.
Bruce Momjian [Mon, 14 Oct 1996 12:12:14 +0000 (12:12 +0000)]
New FAQ's.

28 years agoMore closing of relations left open by parser.
Bruce Momjian [Mon, 14 Oct 1996 03:53:53 +0000 (03:53 +0000)]
More closing of relations left open by parser.

28 years agoChange display of char(5) to (bp)char 5 so psql \d can
Bruce Momjian [Mon, 14 Oct 1996 00:33:47 +0000 (00:33 +0000)]
Change display of char(5) to (bp)char 5 so psql \d can
distinguish between char and char(1).

28 years agoRemoved const warning.
Bruce Momjian [Sun, 13 Oct 1996 18:39:41 +0000 (18:39 +0000)]
Removed const warning.

28 years agoD'Arcy change.
Bruce Momjian [Sun, 13 Oct 1996 18:38:51 +0000 (18:38 +0000)]
D'Arcy change.

28 years agoAdded include needed for recent addition.
Bruce Momjian [Sun, 13 Oct 1996 18:38:04 +0000 (18:38 +0000)]
Added include needed for recent addition.
I hope everyone has sys/socket.h.

28 years agoAdded include needed for recent addition.
Bruce Momjian [Sun, 13 Oct 1996 18:37:19 +0000 (18:37 +0000)]
Added include needed for recent addition.

28 years agoFix structure member name from previous patch.
Bruce Momjian [Sun, 13 Oct 1996 17:13:58 +0000 (17:13 +0000)]
Fix structure member name from previous patch.

28 years agoRemoved duplicate MAXNAMELEN define I accidentally added.
Bruce Momjian [Sun, 13 Oct 1996 13:46:20 +0000 (13:46 +0000)]
Removed duplicate MAXNAMELEN define I accidentally added.

28 years agoAdded consistent include file handling for MAXHOSTNAMELEN.
Bruce Momjian [Sun, 13 Oct 1996 04:50:27 +0000 (04:50 +0000)]
Added consistent include file handling for MAXHOSTNAMELEN.

If I have introduced any syntax errors with this, I will patch them in
the morning, and Marc will have a good laugh.

28 years agoI checked the alter table code, and started suspecting the relation
Bruce Momjian [Sun, 13 Oct 1996 04:26:39 +0000 (04:26 +0000)]
I checked the alter table code, and started suspecting the relation
cache.  I found if I manually added a line to flush the whole relation
cache, the assert error disappeared.  Looking through the code, I found
that the relation cache is flushed at the end of each query if the
reference count is zero for the relation.  However, printf's showed that
the rd_relcnt(reference count) for the accessed query was not returning
to zero after each query.

It turns out the parser was doing a heap_ropen in parser/analyze.c to
get information about the table's columns, but was not doing a
heap_close.

This was causing the query after the ALTER TABLE ADD to see the old
table structure, and the executor's assert was reporting the problem.

28 years agoFix prototypes so postmaster.c will compile.
Bryan Henderson [Sun, 13 Oct 1996 04:01:05 +0000 (04:01 +0000)]
Fix prototypes so postmaster.c will compile.

28 years agoNew host-based authentication -- set up pg_hba.conf instead of pg_hba
Bryan Henderson [Sat, 12 Oct 1996 07:49:56 +0000 (07:49 +0000)]
New host-based authentication -- set up pg_hba.conf instead of pg_hba

28 years agoNew host-based authentication -- send error message when authentication fails
Bryan Henderson [Sat, 12 Oct 1996 07:48:49 +0000 (07:48 +0000)]
New host-based authentication -- send error message when authentication fails

28 years agoNew host-based authentication with ident
Bryan Henderson [Sat, 12 Oct 1996 07:47:12 +0000 (07:47 +0000)]
New host-based authentication with ident

28 years agoNew host-based authentication -- install pg_hba.conf instead of pg_hba
Bryan Henderson [Sat, 12 Oct 1996 07:44:39 +0000 (07:44 +0000)]
New host-based authentication -- install pg_hba.conf instead of pg_hba

28 years agoSeparate general purpose functions from portal functions so they may be
Bryan Henderson [Fri, 11 Oct 1996 09:47:14 +0000 (09:47 +0000)]
Separate general purpose functions from portal functions so they may be
used in test drivers.

28 years agoPrepare for new host-based authentication
Bryan Henderson [Fri, 11 Oct 1996 09:12:18 +0000 (09:12 +0000)]
Prepare for new host-based authentication

28 years agoAdd a MIPS/NetBSD port...
Marc G. Fournier [Fri, 11 Oct 1996 03:56:41 +0000 (03:56 +0000)]
Add a MIPS/NetBSD port...