OSDN Git Service

pg-rex/syncrep.git
23 years agoAdd sys/types include for sockaddr test to configure
Bruce Momjian [Mon, 2 Oct 2000 03:55:15 +0000 (03:55 +0000)]
Add sys/types include for sockaddr test to configure

23 years agoAdd missing python files.
Bruce Momjian [Mon, 2 Oct 2000 03:46:24 +0000 (03:46 +0000)]
Add missing python files.

23 years agoUpdate for PyGreSQL 3.0, from D'Arcy J.M. Cain
Bruce Momjian [Mon, 2 Oct 2000 03:27:33 +0000 (03:27 +0000)]
Update for PyGreSQL 3.0, from D'Arcy J.M. Cain

23 years agoFix quoting bug
Peter Eisentraut [Sun, 1 Oct 2000 22:23:14 +0000 (22:23 +0000)]
Fix quoting bug

23 years agoFix test order
Peter Eisentraut [Sun, 1 Oct 2000 21:19:09 +0000 (21:19 +0000)]
Fix test order

23 years agoFix nasty TRUNCATE bug reported by Darrin Ladd. RelationTruncateIndexes
Tom Lane [Sat, 30 Sep 2000 18:28:53 +0000 (18:28 +0000)]
Fix nasty TRUNCATE bug reported by Darrin Ladd.  RelationTruncateIndexes
would close and then re-open rel being truncated.  Depending on the
luck of the draw, the re-opened relcache entry might or might not be
at the same physical location as before.  Unfortunately, if it wasn't
then heap_truncate would crash and burn, because it still had a pointer
at the old location.  Fix is to open and then close rel in
RelationTruncateIndexes, so that rel's refcount never goes to zero
until heap_truncate is done.

23 years agoConvert more charset/locale documentation to DocBook
Peter Eisentraut [Sat, 30 Sep 2000 16:58:20 +0000 (16:58 +0000)]
Convert more charset/locale documentation to DocBook

23 years agofix shell bug
Peter Eisentraut [Sat, 30 Sep 2000 10:45:17 +0000 (10:45 +0000)]
fix shell bug

23 years agoUpdate FAQ.
Bruce Momjian [Sat, 30 Sep 2000 03:04:39 +0000 (03:04 +0000)]
Update FAQ.

23 years agoUpdate TODO list.
Bruce Momjian [Sat, 30 Sep 2000 02:20:53 +0000 (02:20 +0000)]
Update TODO list.

23 years agoautoconf
Bruce Momjian [Fri, 29 Sep 2000 22:01:56 +0000 (22:01 +0000)]
autoconf

23 years agoPlease apply the following patch to fix problems with the AIX port
Bruce Momjian [Fri, 29 Sep 2000 22:00:49 +0000 (22:00 +0000)]
Please apply the following patch to fix problems with the AIX port
and the fmgr redesign.

It makes the homebrewn dl*() functions for more recent Versions of AIX
obsolete
by using the system dl*() functions instead.

It also fixes the expected file for the horology regression test.
Please regenerate configure from configure.in, I don't have the
environment/time.

Andreas

23 years agoAdd `id' attributes to all `chapter' and `sect1' tags, to generate useful
Peter Eisentraut [Fri, 29 Sep 2000 20:21:34 +0000 (20:21 +0000)]
Add `id' attributes to all `chapter' and `sect1' tags, to generate useful
names for the HTML files (e.g., not x4856.htm).

23 years agoSubselects in FROM clause, per ISO syntax: FROM (SELECT ...) [AS] alias.
Tom Lane [Fri, 29 Sep 2000 18:21:41 +0000 (18:21 +0000)]
Subselects in FROM clause, per ISO syntax: FROM (SELECT ...) [AS] alias.
(Don't forget that an alias is required.)  Views reimplemented as expanding
to subselect-in-FROM.  Grouping, aggregates, DISTINCT in views actually
work now (he says optimistically).  No UNION support in subselects/views
yet, but I have some ideas about that.  Rule-related permissions checking
moved out of rewriter and into executor.
INITDB REQUIRED!

23 years agoNew unified regression test driver, test/regress makefile cleanup,
Peter Eisentraut [Fri, 29 Sep 2000 17:17:41 +0000 (17:17 +0000)]
New unified regression test driver, test/regress makefile cleanup,
add "check" and "installcheck" targets, straighten out make variable naming
of host_os, host_cpu, etc.

23 years agoStandardize on __CYGWIN__ rather than __CYGWIN32__ macro. Doesn't matter
Peter Eisentraut [Fri, 29 Sep 2000 13:53:35 +0000 (13:53 +0000)]
Standardize on __CYGWIN__ rather than __CYGWIN32__ macro.  Doesn't matter
either way (although the former is preferred by the Cygwin folks
themselves), but using only one seems nicer.

23 years agoEnable initdb to create several levels of directories for PGDATA, if
Peter Eisentraut [Fri, 29 Sep 2000 13:38:02 +0000 (13:38 +0000)]
Enable initdb to create several levels of directories for PGDATA, if
`mkdir -p' works.

23 years agoLinking with -lc or -lm crashes on Cygwin; filter out those libraries from
Peter Eisentraut [Fri, 29 Sep 2000 13:36:50 +0000 (13:36 +0000)]
Linking with -lc or -lm crashes on Cygwin; filter out those libraries from
the LIBS variable.

23 years agoCygwin uses _sys_nerr for sys_nerr
Peter Eisentraut [Fri, 29 Sep 2000 13:35:26 +0000 (13:35 +0000)]
Cygwin uses _sys_nerr for sys_nerr

23 years agoBufferAlloc() doesn't allocate write error buffers.
Hiroshi Inoue [Fri, 29 Sep 2000 03:55:45 +0000 (03:55 +0000)]
BufferAlloc() doesn't allocate write error buffers.
Remove compiler waring(my fault).

23 years agoavoid database-wide restart on write error
Hiroshi Inoue [Fri, 29 Sep 2000 01:23:47 +0000 (01:23 +0000)]
avoid database-wide restart on write error

23 years agoCygwin doesn't have DST before 1970 either
Peter Eisentraut [Thu, 28 Sep 2000 08:53:32 +0000 (08:53 +0000)]
Cygwin doesn't have DST before 1970 either

23 years agoFixes for Cygwin, with help from Pete Forman <gsez020@kryten.bedford.waii.com>.
Peter Eisentraut [Wed, 27 Sep 2000 15:17:57 +0000 (15:17 +0000)]
Fixes for Cygwin, with help from Pete Forman <gsez020@kryten.bedford.waii.com>.

Update the installation instructions (formerly misnamed "FAQ"), add configure
checks for some headers rather than having users copy stubs manually (ugh!).
Use Autoconf check for exe extension.  This also avoids inheriting the value
of $(X) from the environment.

23 years agoChange with delimiters to using delimiters in psql manual page.
Bruce Momjian [Wed, 27 Sep 2000 11:56:16 +0000 (11:56 +0000)]
Change with delimiters to using delimiters in psql manual page.

23 years agoReset current user id to session user id during transaction abort
Peter Eisentraut [Wed, 27 Sep 2000 10:41:55 +0000 (10:41 +0000)]
Reset current user id to session user id during transaction abort

23 years ago- Synced preproc.y with gram.y.
Michael Meskes [Tue, 26 Sep 2000 11:41:45 +0000 (11:41 +0000)]
- Synced preproc.y with gram.y.
- Synced keyword.c.
- Added patch by Christof Petig <christof.petig@wtal.de> to fix NOT
  FOUND problem on update/insert/delete.

23 years agoFix compile error in ecpg when enable-multibyte is on. This is due to
Tatsuo Ishii [Tue, 26 Sep 2000 05:42:15 +0000 (05:42 +0000)]
Fix compile error in ecpg when enable-multibyte is on.  This is due to
illegal call to pg_mbclipen() that is for backend only.  However I
have not remove the entire part of the problem, rather mark it with
#ifdef MULTIBYTE_NOTUSED since we should come back with a long range
solution someday.

23 years agodocument configure --without-tk
Peter Eisentraut [Mon, 25 Sep 2000 22:52:49 +0000 (22:52 +0000)]
document configure --without-tk

23 years agoWhen the RI triggers lock the PK table, temporarily switch the current user
Peter Eisentraut [Mon, 25 Sep 2000 22:34:20 +0000 (22:34 +0000)]
When the RI triggers lock the PK table, temporarily switch the current user
id to the owner of the PK table, to avoid permission problems.

23 years agoRevise Tcl/Tk configuration. Make missing Tcl after --with-tcl an error,
Peter Eisentraut [Mon, 25 Sep 2000 22:23:01 +0000 (22:23 +0000)]
Revise Tcl/Tk configuration.  Make missing Tcl after --with-tcl an error,
add --without-tk option to disable Tk.  We don't need the AC_PATH_XTRA
test because tkConfig.sh already contains all the information about how to
compile and link with X.  Also make sure that libpq is up to date for
libpgtcl.  Remove executable bits from pgaccess.sh, but add it to pgaccess.

23 years agoCorrect error in grammar for subselect-in-FROM: SQL spec does not allow
Tom Lane [Mon, 25 Sep 2000 18:38:39 +0000 (18:38 +0000)]
Correct error in grammar for subselect-in-FROM: SQL spec does not allow
omission of alias_clause for a subselect.

23 years agoUse variable aliases, if supplied, rather than real column names in
Tom Lane [Mon, 25 Sep 2000 18:14:55 +0000 (18:14 +0000)]
Use variable aliases, if supplied, rather than real column names in
complaints about ungrouped variables.  This is for consistency with
behavior elsewhere, notably the fact that the relname is reported as
an alias in these same complaints.  Also, it'll work with subselect-
in-FROM where old code didn't.

23 years agoSystem neglected to complain about ungrouped variables passed to
Tom Lane [Mon, 25 Sep 2000 18:09:28 +0000 (18:09 +0000)]
System neglected to complain about ungrouped variables passed to
sublinks when outer query contained aggregates but no GROUP clause.

23 years agoKarel missed an ichar->chr ...
Tom Lane [Mon, 25 Sep 2000 16:36:36 +0000 (16:36 +0000)]
Karel missed an ichar->chr ...

23 years agoChange // to /*.
Bruce Momjian [Mon, 25 Sep 2000 14:40:27 +0000 (14:40 +0000)]
Change // to /*.

23 years ago the patch include:
Bruce Momjian [Mon, 25 Sep 2000 12:58:47 +0000 (12:58 +0000)]
  the patch include:

        - rename ichar() to chr() (discussed with Tom)

        - add docs for oracle compatible routines:

                btrim()
                ascii()
                chr()
                repeat()

        - fix bug with timezone in to_char()

        - all to_char() variants return NULL instead textin("")
          if it's needful.

 The contrib/odbc is without changes and contains same routines as main
tree ... because I not sure how plans are Thomas with this :-)

                                        Karel
---------------------------------------------------------------------------

This effectively one line patch should fix the fact that
foreign key definitions in create table were erroring if
a primary key was defined.  I was using the columns
list to get the columns of the table for comparison, but
it got reused as a temporary list inside the primary key
stuff.

Stephan Szabo

23 years agodisable plperl again.
Bruce Momjian [Mon, 25 Sep 2000 12:43:24 +0000 (12:43 +0000)]
disable plperl again.

23 years agoAcquire read lock on a buffer while writing it out, to prevent
Tom Lane [Mon, 25 Sep 2000 04:11:09 +0000 (04:11 +0000)]
Acquire read lock on a buffer while writing it out, to prevent
concurrent modifications to the page by other backends.

23 years agoTweak resultmap so that geometry regress test selects the right comparison
Tom Lane [Sun, 24 Sep 2000 04:07:11 +0000 (04:07 +0000)]
Tweak resultmap so that geometry regress test selects the right comparison
file on LinuxPPC, which is identified by config.guess as
powerpc-unknown-linux-gnu (not ...-gnulibc1).

23 years agoClean up some ugly coding (hardwired constants) in index_formtuple.
Tom Lane [Sat, 23 Sep 2000 22:40:12 +0000 (22:40 +0000)]
Clean up some ugly coding (hardwired constants) in index_formtuple.

23 years agoThese are obsolete.
Peter Eisentraut [Sat, 23 Sep 2000 00:42:36 +0000 (00:42 +0000)]
These are obsolete.

23 years agoDon't free() the result of getpwuid().
Peter Eisentraut [Fri, 22 Sep 2000 23:02:00 +0000 (23:02 +0000)]
Don't free() the result of getpwuid().
Discovered by Jason Tishler <Jason.Tishler@dothill.com>.

23 years agoSetPGVariable() crashed on SET DateStyle TO DEFAULT and related cases.
Tom Lane [Fri, 22 Sep 2000 15:34:31 +0000 (15:34 +0000)]
SetPGVariable() crashed on SET DateStyle TO DEFAULT and related cases.

23 years agoAdd check of 'SET DateStyle TO DEFAULT'.
Tom Lane [Fri, 22 Sep 2000 15:33:31 +0000 (15:33 +0000)]
Add check of 'SET DateStyle TO DEFAULT'.

23 years agoReplace brain-dead Autoconf macros AC_ARG_{ENABLE,WITH} with something
Peter Eisentraut [Thu, 21 Sep 2000 20:17:43 +0000 (20:17 +0000)]
Replace brain-dead Autoconf macros AC_ARG_{ENABLE,WITH} with something
that's actually useful, robust, consistent.

Better plan to generate aclocal.m4 as well: use m4 include directives,
rather than cat.

23 years ago*** empty log message ***
Michael Meskes [Thu, 21 Sep 2000 11:56:08 +0000 (11:56 +0000)]
*** empty log message ***

23 years agoFix coredump in copyCommentStmt().
Tom Lane [Wed, 20 Sep 2000 15:28:01 +0000 (15:28 +0000)]
Fix coredump in copyCommentStmt().

23 years ago*** empty log message ***
Michael Meskes [Wed, 20 Sep 2000 13:25:52 +0000 (13:25 +0000)]
*** empty log message ***

23 years agoMake sure that FlushRelationBuffers() is invoked by all paths through
Tom Lane [Tue, 19 Sep 2000 19:30:03 +0000 (19:30 +0000)]
Make sure that FlushRelationBuffers() is invoked by all paths through
vacuum.c.  This is needed to make the world safe for pg_upgrade.

23 years agoFix GEQO optimizer to work correctly with new outer-join-capable
Tom Lane [Tue, 19 Sep 2000 18:42:34 +0000 (18:42 +0000)]
Fix GEQO optimizer to work correctly with new outer-join-capable
query representation.  Note that GEQO_RELS setting is now interpreted
as the number of top-level items in the FROM list, not necessarily the
number of relations in the query.  This seems appropriate since we are
only doing join-path searching over the top-level items.

23 years agoImplement differentiation between CURRENT_USER and SESSION_USER as per SQL.
Peter Eisentraut [Tue, 19 Sep 2000 18:18:04 +0000 (18:18 +0000)]
Implement differentiation between CURRENT_USER and SESSION_USER as per SQL.
There is still no effective difference but it will kick in once setuid
functions exist (not included here).  Make old getpgusername() alias for
current_user.

23 years ago*** empty log message ***
Michael Meskes [Tue, 19 Sep 2000 11:47:16 +0000 (11:47 +0000)]
*** empty log message ***

23 years agoMake rule lister use aliases from FROM clause when a table column has
Tom Lane [Mon, 18 Sep 2000 20:14:23 +0000 (20:14 +0000)]
Make rule lister use aliases from FROM clause when a table column has
been given an alias.  Otherwise, results are incorrect.

23 years agoMove contrib/odbc/odbc.sql together with odbc driver, remove old files;
Peter Eisentraut [Mon, 18 Sep 2000 20:11:37 +0000 (20:11 +0000)]
Move contrib/odbc/odbc.sql together with odbc driver, remove old files;
revise ODBC "installation instructions"; update some other outdated
documentation; update contrib documentation

23 years agoOnly retrieve view definition if relation is a view (case statement)
Philip Warner [Mon, 18 Sep 2000 06:47:46 +0000 (06:47 +0000)]
Only retrieve view definition if relation is a view (case statement)

23 years agoUses column select expressions to get object object owners and tests for blank names
Philip Warner [Mon, 18 Sep 2000 03:24:03 +0000 (03:24 +0000)]
Uses column select expressions to get object object owners and tests for blank names

23 years agoParse JOIN/ON conditions with the proper visibility of input columns,
Tom Lane [Sun, 17 Sep 2000 22:21:27 +0000 (22:21 +0000)]
Parse JOIN/ON conditions with the proper visibility of input columns,
ie, consider only the columns coming from the JOIN clause's sub-clauses.
Also detect attempts to reference columns belonging to other tables
(which would still be possible using an explicitly-qualified name).
I'm not sure this implements the spec's semantics 100% accurately, but
at least it gives plausible behavior.

23 years agoAdd a blurb about what tools are required to build.
Peter Eisentraut [Sun, 17 Sep 2000 20:35:27 +0000 (20:35 +0000)]
Add a blurb about what tools are required to build.

23 years agopsql forgot to close connection before re-issuing password prompt.
Peter Eisentraut [Sun, 17 Sep 2000 20:33:45 +0000 (20:33 +0000)]
psql forgot to close connection before re-issuing password prompt.
([BUGS] psql can crash the backend on login, 2000-09-03)

23 years agoImprove error message for dumpRules() failure
Tom Lane [Sun, 17 Sep 2000 20:01:28 +0000 (20:01 +0000)]
Improve error message for dumpRules() failure

23 years agoSupport for DESTDIR make variable. This is used as in `make install
Peter Eisentraut [Sun, 17 Sep 2000 13:02:52 +0000 (13:02 +0000)]
Support for DESTDIR make variable. This is used as in `make install
DESTDIR=/else/where' and prepends the value of DESTDIR to the full
installation paths (e.g., /else/where/usr/local/pgsql/bin). This allows
users to install the package into a location different from the one that
was configured and hard-coded into various scripts, e.g., for creating
binary packages.

DESTDIR is in many cases preferrable over `make install
prefix=/else/where' because

a) `prefix' affects the path that is hard-coded into the files, which can
lead to a `make install prefix=xxx' (as done by the regression test
driver) corrupting the files in the source tree with wrong paths.

b) it doesn't work at all if a directory was overridden to not depend on
`prefix', e.g., --sysconfdir=/etc.

(Updating the regression test driver to use DESTDIR is a separate
undertaking.)

See also autoconf@gnu.org, From: Akim Demaille <akim@epita.fr>, Date: 08
Sep 2000 12:48:59 +0200, Message-ID:
<mv4em2vb1lw.fsf@nostromo.lrde.epita.fr>, Subject: Re: HTML format
documentation.

23 years agoAdd ILIKE operators ~~*, !~~* to documentation.
Tom Lane [Fri, 15 Sep 2000 20:20:11 +0000 (20:20 +0000)]
Add ILIKE operators ~~*, !~~* to documentation.

23 years agoVariant horology resultfiles are inconsistent with the test, again.
Tom Lane [Fri, 15 Sep 2000 19:37:49 +0000 (19:37 +0000)]
Variant horology resultfiles are inconsistent with the test, again.

23 years agoReimplement LIKE/ESCAPE as operators so that indexscan optimization
Tom Lane [Fri, 15 Sep 2000 18:45:31 +0000 (18:45 +0000)]
Reimplement LIKE/ESCAPE as operators so that indexscan optimization
can still work, per recent discussion on pghackers.  Correct some bugs
in ILIKE implementation.

23 years agoAllow PL/pgSQL accept non ascii identifiers
Tatsuo Ishii [Fri, 15 Sep 2000 11:59:40 +0000 (11:59 +0000)]
Allow PL/pgSQL accept non ascii identifiers

23 years agoMinor code cleanup (cast away const-ness)
Philip Warner [Fri, 15 Sep 2000 04:57:09 +0000 (04:57 +0000)]
Minor code cleanup (cast away const-ness)

23 years ago- Support for relkind = RELKIND_VIEW.
Philip Warner [Fri, 15 Sep 2000 04:35:16 +0000 (04:35 +0000)]
- Support for relkind = RELKIND_VIEW.
- Use symbols for tests on relkind (ie. use RELKIND_VIEW, not 'v')
- Fix bug in support for -b option (== --blobs).
- Dump views as views (using 'create view').
- Remove 'isViewRule' since we check the relkind when getting tables.
- Now uses temp table 'pgdump_oid' rather than 'pg_dump_oid' (errors otherwise).
- Added extra param for specifying handling of OID=0 and which typename to output.
- Fixed bug in SQL scanner when SQL contained braces. (in rules)
- Use format_type function wherever possible

23 years agoMove a few specific tests to isolate homogenous type tests into
Thomas G. Lockhart [Thu, 14 Sep 2000 15:55:07 +0000 (15:55 +0000)]
Move a few specific tests to isolate homogenous type tests into
 regression tests for specific types, and move a few others to the
 cross-type "horology" test.
Rearrange the test order slightly, and move the abstime test to the
 "parallel safe" area.
Hand-patch the results for "1947" and for "solaris", so those may not
 be exactly correct.

23 years agoAdd new FAQ's.
Bruce Momjian [Thu, 14 Sep 2000 02:21:31 +0000 (02:21 +0000)]
Add new FAQ's.

23 years agorename README.Charsets
Bruce Momjian [Thu, 14 Sep 2000 02:17:28 +0000 (02:17 +0000)]
rename README.Charsets

23 years agotypo fix
Bruce Momjian [Tue, 12 Sep 2000 21:12:45 +0000 (21:12 +0000)]
typo fix

23 years agoFirst cut at full support for OUTER JOINs. There are still a few loose
Tom Lane [Tue, 12 Sep 2000 21:07:18 +0000 (21:07 +0000)]
First cut at full support for OUTER JOINs.  There are still a few loose
ends to clean up (see my message of same date to pghackers), but mostly
it works.  INITDB REQUIRED!

23 years agotypo fix
Bruce Momjian [Tue, 12 Sep 2000 20:58:14 +0000 (20:58 +0000)]
typo fix

23 years agodatatype -> data type.
Bruce Momjian [Tue, 12 Sep 2000 20:52:08 +0000 (20:52 +0000)]
datatype -> data type.

23 years agoFix relation-to-view conversion so that it doesn't try to convert a plain
Tom Lane [Tue, 12 Sep 2000 20:38:09 +0000 (20:38 +0000)]
Fix relation-to-view conversion so that it doesn't try to convert a plain
relation to a view when you create an ON INSERT/UPDATE/DELETE rule ...

23 years agoUpdate expected/horology-no-DST-before-1970.out for new horology test.
Tom Lane [Tue, 12 Sep 2000 20:11:23 +0000 (20:11 +0000)]
Update expected/horology-no-DST-before-1970.out for new horology test.
There are still two out-of-date variants for horology; shame shame.

23 years agoClean up compiler warning.
Tom Lane [Tue, 12 Sep 2000 19:41:40 +0000 (19:41 +0000)]
Clean up compiler warning.

23 years agoSuppress compile warnings caused by failure to use PG_RETURN macro.
Tom Lane [Tue, 12 Sep 2000 19:39:42 +0000 (19:39 +0000)]
Suppress compile warnings caused by failure to use PG_RETURN macro.

23 years agoAs if my JDBC patch hasn't already caused enough grief, there is now a
Bruce Momjian [Tue, 12 Sep 2000 18:56:04 +0000 (18:56 +0000)]
As if my JDBC patch hasn't already caused enough grief, there is now a
one-line change necessary. Due to the Mark Holloman "New Relkind for
Views" patch, my support for views in the driver will need to be updated
to match. The change to DatabaseMetaData.getTableTypes[][] is as
follows:

-    {"VIEW",           "(relkind='r' and relhasrules='t' and relname !~
'^pg_' and relname !~ '^xinv')"},
+    {"VIEW",           "(relkind='v' and relname !~ '^pg_' and relname
!~ '^xinv')"},

Christopher Cain

23 years agoRemove incorrect extra heap_close when deleting a builtin function.
Tom Lane [Tue, 12 Sep 2000 16:48:55 +0000 (16:48 +0000)]
Remove incorrect extra heap_close when deleting a builtin function.

23 years agoUpdate FAQ.
Bruce Momjian [Tue, 12 Sep 2000 15:47:20 +0000 (15:47 +0000)]
Update FAQ.

23 years agoBtree WAL records.
Vadim B. Mikheev [Tue, 12 Sep 2000 06:07:52 +0000 (06:07 +0000)]
Btree WAL records.

23 years agoAdd regression tests for date, time, and time with time zone types.
Thomas G. Lockhart [Tue, 12 Sep 2000 05:42:20 +0000 (05:42 +0000)]
Add regression tests for date, time, and time with time zone types.
Modify date->timestamp conversion to use mktime().
 This should do better than before around Daylight Savings Time
 transitions.

23 years agoModify date->timestamp conversion to use mktime().
Thomas G. Lockhart [Tue, 12 Sep 2000 05:41:37 +0000 (05:41 +0000)]
Modify date->timestamp conversion to use mktime().
 This should do better than before around Daylight Savings Time
 transitions.

23 years agoConsolidate FAQ and README info on character sets into new chapter.
Thomas G. Lockhart [Tue, 12 Sep 2000 05:37:09 +0000 (05:37 +0000)]
Consolidate FAQ and README info on character sets into new chapter.

23 years agoUpdate to refer to more recent stylesheets in my home directory.
Thomas G. Lockhart [Tue, 12 Sep 2000 05:36:06 +0000 (05:36 +0000)]
Update to refer to more recent stylesheets in my home directory.
 These stylesheets are pretty old, but the 15x series crashes FreeBSD jade.

23 years agoClean up markup to be more correct.
Thomas G. Lockhart [Tue, 12 Sep 2000 05:34:28 +0000 (05:34 +0000)]
Clean up markup to be more correct.
 From "He Weiping (Laser Henry)" <laser@zhengmai.com.cn>.

23 years agoFix inconsistancies in field names. From Tom Samplonius <tom@sdf.com>.
Thomas G. Lockhart [Tue, 12 Sep 2000 05:30:06 +0000 (05:30 +0000)]
Fix inconsistancies in field names. From Tom Samplonius <tom@sdf.com>.

23 years agoAttached is a patch that prevents a NullPointerException in the JDBC
Bruce Momjian [Tue, 12 Sep 2000 05:18:25 +0000 (05:18 +0000)]
Attached is a patch that prevents a NullPointerException in the JDBC
driver if the translations files have not been properly installed.  (We
carefully avoided installing the translations file in a controlled
environment here specifically to test for such a bug. :-)

See attached description for more details.

William
--
William Webber                               william@peopleweb.net.au

23 years agoThis patch implements the following command:
Bruce Momjian [Tue, 12 Sep 2000 05:09:57 +0000 (05:09 +0000)]
This patch implements the following command:

ALTER TABLE <tablename> OWNER TO <username>

Only a superuser may execute the command.

--
Mark Hollomon
mhh@mindspring.com

23 years agoAttached are a patch to allow the charset encoding used by the JDBC
Bruce Momjian [Tue, 12 Sep 2000 04:58:50 +0000 (04:58 +0000)]
Attached are a patch to allow the charset encoding used by the JDBC
driver to be set, and a description of said patch.  Please refer to
the latter for more information.

William
--
William Webber                               william@peopleweb.net.au

23 years agoApplied to jdbc1 and jdbc2.
Bruce Momjian [Tue, 12 Sep 2000 04:51:43 +0000 (04:51 +0000)]
Applied to jdbc1 and jdbc2.

This is a patch which lets the DatabaseMetaData return the object type
when getTables(....) is called.  It does not really fix any bug, but it
fills in some functionality that should be there anyway.  The diff
included here is off of the CVS as of just now :)

----------------------------------------------------------------
Travis Bauer | CS Grad Student | IU |www.cs.indiana.edu/~trbauer
----------------------------------------------------------------

23 years agoThis patch implements a different "relkind"
Bruce Momjian [Tue, 12 Sep 2000 04:49:17 +0000 (04:49 +0000)]
This patch implements a different "relkind"
for views. Views are now have a "relkind" of
RELKIND_VIEW instead of RELKIND_RELATION.

Also, views no longer have actual heap storage
files.

The following changes were made

1. CREATE VIEW sets the new relkind

2. The executor complains if a DELETE or
        INSERT references a view.

3. DROP RULE complains if an attempt is made
        to delete a view SELECT rule.

4. CREATE RULE "_RETmytable" AS ON SELECT TO mytable DO INSTEAD ...
        1. checks to make sure mytable is empty.
        2. sets the relkind to RELKIND_VIEW.
        3. deletes the heap storage files.
5. LOCK myview is not allowed. :)

6. the regression test type_sanity was changed to
        account for the new relkind value.

7. CREATE INDEX ON myview ... is not allowed.

8. VACUUM myview is not allowed.
        VACUUM automatically skips views when do the entire
        database.

9. TRUNCATE myview is not allowed.

THINGS LEFT TO THINK ABOUT

o pg_views

o pg_dump

o pgsql (\d \dv)
o Do we really want to be able to inherit from views?

o Is 'DROP TABLE myview' OK?

--
Mark Hollomon

23 years agoautoconf
Bruce Momjian [Tue, 12 Sep 2000 04:43:21 +0000 (04:43 +0000)]
autoconf

23 years agoAttached is a small patch that fixes a small "bug" with specifying the
Bruce Momjian [Tue, 12 Sep 2000 04:41:17 +0000 (04:41 +0000)]
Attached is a small patch that fixes a small "bug" with specifying the
souce directory of OpenSSL as argument to the --with-openssl - argument.

./configure --with-openssl=/tmp/openssl-0.9.5/ - the libcrypto.a is in
the
"base" directory, not lib/.

Also, there's 2-3 typo fixes to the documentation for SSL - it should be
"server.crt", not "server.cert".

Dominic J. Eidson

23 years agoPrevious patch backed out.
Bruce Momjian [Tue, 12 Sep 2000 04:33:18 +0000 (04:33 +0000)]
Previous patch backed out.

Here is a patch against CVS (without my earlier patch)
to disallow

LOCK x

if x is a view.

It does not use the SPI interface.

--
Mark Hollomon

23 years agoThis patch is for the TODO item
Bruce Momjian [Tue, 12 Sep 2000 04:30:08 +0000 (04:30 +0000)]
This patch is for the TODO item

* Disallow LOCK on view

src/backend/commands/command.c is the only affected file

--
Mark Hollomon

23 years agoThis patch, when applied in src/pl will unbreak plperl in
Bruce Momjian [Tue, 12 Sep 2000 04:28:30 +0000 (04:28 +0000)]
This patch, when applied in src/pl will unbreak plperl in
7.0.2 release. Sorry, if that's fixed ages ago - I don't track
development versions of PostgreSQL.

Patch is just a little bit tested (some valid functions created and
successfully run as well as some erroneous ones created and emitted proper
error messages when used).

My platform is FreeBSD 5.0-CURRENT (with perl 5.6.0 provided in the
base system).

Alex Kapranoff

23 years agoThis patch for the 7.0.2 JDBC interface addresses four issues I
Bruce Momjian [Tue, 12 Sep 2000 04:16:59 +0000 (04:16 +0000)]
This patch for the 7.0.2 JDBC interface addresses four issues I
encountered while getting my reporting tool up and running with the
driver. All changes are in the DatabaseMetaData class.

Problem: The getDatabaseProductVersion() method was returning "6.5.2"
Resolution: Changed it to return "7.0.2"

Problem: A call to getTables() with an unsupported table type (in the
String array) resulted in a malformed SQL statement and subsequent
parsing error
Resolution: Unsupported table types are now ignored without error

Problem: In a getTables() call, tables and views were both returned by
the "TABLE" table type, and the "VIEW" table type was unsupported
Resolution: Changed the "TABLE" type to return only physical tables and
added support for the "VIEW" table type (returning only views)

Problem: The getIdentifierQuoteString() method was returning null
Resolution: This method now returns a double-quote

Christopher Cain

23 years agoO.K. -
Bruce Momjian [Tue, 12 Sep 2000 04:15:58 +0000 (04:15 +0000)]
O.K. -
Here's the multibyte aware version of my patch to fix the truncation
of the rulename autogenerated during a CREATE VIEW. I've modified all
the places in the backend that want to construct the rulename to use
the MakeRetrieveViewRuleName(), where I put the #ifdef MULTIBYTE, so
that's the only place that knows how to construct a view rulename. Except
pg_dump, where I replicated the code, since it's a standalone binary.

The only effect the enduser will see is that views with names len(name)
> NAMEDATALEN-4 will fail to be created, if the derived rulename clases
with an existing rule: i.e. the user is trying to create two views with
long names whose first difference is past NAMEDATALEN-4 (but before
NAMEDATALEN: that'll error out after the viewname truncation.) In no
case will the user get left with a table without a view rule, as the
current code does.

Ross Reedstrom