]> granicus.if.org Git - postgresql/log
postgresql
20 years agoWhen testing whether a sub-plan can do projection, use a general-purpose
Tom Lane [Sun, 18 Jan 2004 00:50:03 +0000 (00:50 +0000)]
When testing whether a sub-plan can do projection, use a general-purpose
check instead of hardwiring assumptions that only certain plan node types
can appear at the places where we are testing.  This was always a pretty
fragile assumption, and it turns out to be broken in 7.4 for certain cases
involving IN-subselect tests that need type coercion.
Also, modify code that builds finished Plan tree so that node types that
don't do projection always copy their input node's targetlist, rather than
having the tlist passed in from the caller.  The old method makes it too
easy to write broken code that thinks it can modify the tlist when it
cannot.

20 years agoRepair misestimation of indexscan CPU costs. When an indexqual contains
Tom Lane [Sat, 17 Jan 2004 20:09:35 +0000 (20:09 +0000)]
Repair misestimation of indexscan CPU costs.  When an indexqual contains
a run-time key (that is, a nonconstant expression compared to the index
variable), the key is evaluated just once per scan, but we were charging
costs as though it were evaluated once per visited index entry.

20 years agoTighten short-circuit tests for deciding whether we need to invoke
Tom Lane [Fri, 16 Jan 2004 20:51:30 +0000 (20:51 +0000)]
Tighten short-circuit tests for deciding whether we need to invoke
tuptoaster.c --- fields that are compressed in-line are not a reason
to invoke the toaster.  Along the way, add a couple more htup.h macros
to eliminate confusing negated tests, and get rid of the already
vestigial TUPLE_TOASTER_ACTIVE symbol.

20 years agoSome refinement of XSLT processing: The result now looks just about as good
Peter Eisentraut [Thu, 15 Jan 2004 23:51:32 +0000 (23:51 +0000)]
Some refinement of XSLT processing: The result now looks just about as good
as the result of the SGML-based tool chain.

20 years agoAdjusted calculation of shared memory requirements to new
Jan Wieck [Thu, 15 Jan 2004 16:14:26 +0000 (16:14 +0000)]
Adjusted calculation of shared memory requirements to new
ARC buffer replacement strategy.

Jan

20 years agoFix permission-checking bug reported by Tim Burgess 10-Feb-03 (this time
Tom Lane [Wed, 14 Jan 2004 23:01:55 +0000 (23:01 +0000)]
Fix permission-checking bug reported by Tim Burgess 10-Feb-03 (this time
for sure...).  Rather than relying on the query context of a rangetable
entry to identify what permissions it wants checked, store a full AclMode
mask in each RTE, and check exactly those bits.  This allows an RTE
specifying, say, INSERT privilege on a view to be copied into a derived
UPDATE query without changing meaning.  Per recent discussion thread.
initdb forced due to change of stored rule representation.

20 years agoTranslation updates
Dennis Bjorklund [Wed, 14 Jan 2004 12:14:01 +0000 (12:14 +0000)]
Translation updates

20 years agopg_settings should have UPDATE privilege allowed to public.
Tom Lane [Wed, 14 Jan 2004 03:46:28 +0000 (03:46 +0000)]
pg_settings should have UPDATE privilege allowed to public.

20 years agoThe no-updates-to-system-catalogs-unless-usecatupd restriction should
Tom Lane [Wed, 14 Jan 2004 03:44:53 +0000 (03:44 +0000)]
The no-updates-to-system-catalogs-unless-usecatupd restriction should
not apply to system views.  It never mattered before 7.4, but it does now.

20 years agoRevert ill-starred change of 13-Feb-02: it appeared to fix a problem of
Tom Lane [Wed, 14 Jan 2004 03:39:22 +0000 (03:39 +0000)]
Revert ill-starred change of 13-Feb-02: it appeared to fix a problem of
incorrect permissions checking, but in fact disabled most all permissions
checks for view updates.  This corrects problems reported by Sergey
Yatskevich among others, at the cost of re-introducing the problem
previously reported by Tim Burgess.  However, since we'd lived with that
problem for quite awhile without knowing it, we can live with it awhile
longer until a proper fix can be made in 7.5.

20 years agoTranslation update
Peter Eisentraut [Tue, 13 Jan 2004 18:58:41 +0000 (18:58 +0000)]
Translation update

20 years agoMake rule and stylesheet to try out XSLT for DocBook processing
Peter Eisentraut [Tue, 13 Jan 2004 18:45:19 +0000 (18:45 +0000)]
Make rule and stylesheet to try out XSLT for DocBook processing

20 years agoFix ecpg test program to properly access int* null indicator.
Bruce Momjian [Tue, 13 Jan 2004 01:32:52 +0000 (01:32 +0000)]
Fix ecpg test program to properly access int* null indicator.

20 years agoFix another place that wasn't maintaining AND/OR flatness of an
Tom Lane [Mon, 12 Jan 2004 22:20:28 +0000 (22:20 +0000)]
Fix another place that wasn't maintaining AND/OR flatness of an
already-canonicalized qual expression.

20 years agoPreserve AND/OR flatness during eval_const_expressions(). This seems a
Tom Lane [Mon, 12 Jan 2004 20:48:15 +0000 (20:48 +0000)]
Preserve AND/OR flatness during eval_const_expressions().  This seems a
useful improvement in any case, and it keeps the new logic for restrictinfo
structures happy.  Per report from Kris Jurka.

20 years agoTranslation updates
Dennis Bjorklund [Sun, 11 Jan 2004 19:42:44 +0000 (19:42 +0000)]
Translation updates

20 years agoMark string for translation.
Dennis Bjorklund [Sun, 11 Jan 2004 19:10:49 +0000 (19:10 +0000)]
Mark string for translation.

20 years agoPut See Also links back in alphabetical order.
Peter Eisentraut [Sun, 11 Jan 2004 09:24:17 +0000 (09:24 +0000)]
Put See Also links back in alphabetical order.

20 years agoTranslation updates
Dennis Bjorklund [Sun, 11 Jan 2004 06:55:10 +0000 (06:55 +0000)]
Translation updates

20 years agoTranslation updates
Dennis Bjorklund [Sun, 11 Jan 2004 06:50:27 +0000 (06:50 +0000)]
Translation updates

20 years agoMinor documentation improvements.
Neil Conway [Sun, 11 Jan 2004 05:46:58 +0000 (05:46 +0000)]
Minor documentation improvements.

20 years agoAdd CREATE TRIGGER, CREATE INDEX, and CREATE SEQUENCE to the list of
Neil Conway [Sun, 11 Jan 2004 04:58:17 +0000 (04:58 +0000)]
Add CREATE TRIGGER, CREATE INDEX, and CREATE SEQUENCE to the list of
expressions supported by CREATE SCHEMA.

Also added the beginning of some regression tests for CREATE SCHEMA;
plenty more work is needed here.

20 years agoDrops in the CreateProcess calls for Win32 (essentially wrapping up the
Bruce Momjian [Sun, 11 Jan 2004 03:49:31 +0000 (03:49 +0000)]
Drops in the CreateProcess calls for Win32 (essentially wrapping up the
fork/exec portion of the port), and fixes a handful of whitespace issues

Claudio Natoli

20 years agoRemove the following unused type definitions: bool8, bool16, bool32, word8,
Neil Conway [Sat, 10 Jan 2004 23:39:51 +0000 (23:39 +0000)]
Remove the following unused type definitions: bool8, bool16, bool32, word8,
word16, and word32.

20 years agoImplement "WITH / WITHOID OIDS" clause for CREATE TABLE AS. This is
Neil Conway [Sat, 10 Jan 2004 23:28:45 +0000 (23:28 +0000)]
Implement "WITH / WITHOID OIDS" clause for CREATE TABLE AS. This is
intended to allow application authors to insulate themselves from
changes to the default value of 'default_with_oids' in future releases
of PostgreSQL.

This patch also fixes a bug in the earlier implementation of the
'default_with_oids' GUC variable: code in gram.y should not examine
the value of GUC variables directly due to synchronization issues.

20 years agoTranslation updates
Peter Eisentraut [Sat, 10 Jan 2004 22:18:28 +0000 (22:18 +0000)]
Translation updates

20 years agoImprove has_nullable_targetlist() to allow strict functions of simple
Tom Lane [Sat, 10 Jan 2004 18:13:53 +0000 (18:13 +0000)]
Improve has_nullable_targetlist() to allow strict functions of simple
variables, not just simple variables.  This was foreseen in the original
coding of this routine, but not implemented until now.  Responds to
performance gripe from Laurent Perez.

20 years agoTranslation updates
Dennis Bjorklund [Sat, 10 Jan 2004 15:02:50 +0000 (15:02 +0000)]
Translation updates

20 years agoTranslation updates
Dennis Bjorklund [Sat, 10 Jan 2004 10:39:26 +0000 (10:39 +0000)]
Translation updates

20 years agoTranslation updates
Dennis Bjorklund [Sat, 10 Jan 2004 09:09:40 +0000 (09:09 +0000)]
Translation updates

20 years agoMake non-wait pg_ctl start/stop message more accurate.
Bruce Momjian [Sat, 10 Jan 2004 02:55:14 +0000 (02:55 +0000)]
Make non-wait pg_ctl start/stop message more accurate.

20 years agoAllow BEGIN WORK to specify transaction isolation level, like START
Bruce Momjian [Sat, 10 Jan 2004 02:21:08 +0000 (02:21 +0000)]
Allow BEGIN WORK to specify transaction isolation level, like START
TRANSACTION.

20 years agoFix subquery pullup logic to not be fooled when a view that appears
Tom Lane [Sat, 10 Jan 2004 00:30:21 +0000 (00:30 +0000)]
Fix subquery pullup logic to not be fooled when a view that appears
'simple' references another view that is not simple.  Must recheck
conditions after performing recursive pullup.  Per example from
Laurent Perez, 9-Jan-04.

20 years agoOops, can't do cpp tests inside macros; make two macros instead.
Bruce Momjian [Fri, 9 Jan 2004 23:29:31 +0000 (23:29 +0000)]
Oops, can't do cpp tests inside macros; make two macros instead.

20 years agoRemove ereport() from failed connection; already done, per Tom.
Bruce Momjian [Fri, 9 Jan 2004 23:27:20 +0000 (23:27 +0000)]
Remove ereport() from failed connection; already done, per Tom.

20 years agoFix for cancel key restarting postmaster, from Claudio
Bruce Momjian [Fri, 9 Jan 2004 23:11:39 +0000 (23:11 +0000)]
Fix for cancel key restarting postmaster, from Claudio

20 years agoHave psql help show \w parameter as requied.
Bruce Momjian [Fri, 9 Jan 2004 21:15:51 +0000 (21:15 +0000)]
Have psql help show \w parameter as requied.

20 years agoDone:
Bruce Momjian [Fri, 9 Jan 2004 21:12:55 +0000 (21:12 +0000)]
Done:
> * -Have psql '\i ~/<tab><tab>' actually load files it displays from home dir

20 years agoAllow psql to handle tilde user expansion for file names.
Bruce Momjian [Fri, 9 Jan 2004 21:12:20 +0000 (21:12 +0000)]
Allow psql to handle tilde user expansion for file names.

Zach Irmen

20 years agoChange PG_DELAY from msec to usec and use it consistenly rather than
Bruce Momjian [Fri, 9 Jan 2004 21:08:50 +0000 (21:08 +0000)]
Change PG_DELAY from msec to usec and use it consistenly rather than
select().   Add Win32 Sleep() for delay.

20 years agoTranslation updates
Dennis Bjorklund [Fri, 9 Jan 2004 20:36:00 +0000 (20:36 +0000)]
Translation updates

20 years agoAdd WIN32 pipe implementation that uses sockets.
Bruce Momjian [Fri, 9 Jan 2004 04:58:09 +0000 (04:58 +0000)]
Add WIN32 pipe implementation that uses sockets.

Claudio Natoli

20 years agoRemove duplicate include of value.h. I think it got in because Tom and I both added it.
Bruce Momjian [Fri, 9 Jan 2004 03:07:32 +0000 (03:07 +0000)]
Remove duplicate include of value.h.  I think it got in because Tom and I both added it.

20 years agoAdd comment about synchronous signals.
Bruce Momjian [Fri, 9 Jan 2004 02:17:15 +0000 (02:17 +0000)]
Add comment about synchronous signals.

20 years agoAllow libpq to do thread-safe SIGPIPE handling. This allows it to
Bruce Momjian [Fri, 9 Jan 2004 02:02:43 +0000 (02:02 +0000)]
Allow libpq to do thread-safe SIGPIPE handling.  This allows it to
ignore SIGPIPE from send() in libpq, but terminate on any other SIGPIPE,
unless the user installs their own signal handler.

This is a minor fix because the only time you get SIGPIPE from libpq's
send() is when the backend dies.

20 years agoGerman FAQ update.
Bruce Momjian [Fri, 9 Jan 2004 01:51:31 +0000 (01:51 +0000)]
German FAQ update.

Ian Barwick

20 years agoFix incorrect SQL syntax emitted when -E is given without -P.
Tom Lane [Fri, 9 Jan 2004 00:15:11 +0000 (00:15 +0000)]
Fix incorrect SQL syntax emitted when -E is given without -P.
Report and fix from Martin Pitt.

20 years agoRemove broken (and unnecessary) definition of operator <> for _int4
Tom Lane [Thu, 8 Jan 2004 23:40:27 +0000 (23:40 +0000)]
Remove broken (and unnecessary) definition of operator <> for _int4
datatype; the generic array comparators added in 7.4 supersede this.
Per report and patch from Korea PostgreSQL Users' Group.

20 years agoFix a spelling mistake in a comment.
Neil Conway [Thu, 8 Jan 2004 20:04:41 +0000 (20:04 +0000)]
Fix a spelling mistake in a comment.

20 years agoThrow compile error if our non-thread-safe snprintf is used.
Bruce Momjian [Thu, 8 Jan 2004 17:15:54 +0000 (17:15 +0000)]
Throw compile error if our non-thread-safe snprintf is used.

20 years agoUpdate Russian FAQ.
Bruce Momjian [Thu, 8 Jan 2004 16:01:58 +0000 (16:01 +0000)]
Update Russian FAQ.

Viktor Vislobokov

20 years agoUpdate OSF thread flags for non-gcc compiles.
Bruce Momjian [Thu, 8 Jan 2004 15:51:24 +0000 (15:51 +0000)]
Update OSF thread flags for non-gcc compiles.

20 years agoTranslation updates
Dennis Bjorklund [Thu, 8 Jan 2004 10:47:40 +0000 (10:47 +0000)]
Translation updates

20 years agoTranslation updates
Dennis Bjorklund [Thu, 8 Jan 2004 08:36:50 +0000 (08:36 +0000)]
Translation updates

20 years ago[Forced commit to add log message that I fat-fingered last time]
Tom Lane [Thu, 8 Jan 2004 06:01:21 +0000 (06:01 +0000)]
[Forced commit to add log message that I fat-fingered last time]

Give a more reasonable error message when lock file exists but has
zero length; prior code confused this with could-not-read-file case.

20 years agoFix Solaris thread build settings for non-gcc compilers.
Bruce Momjian [Thu, 8 Jan 2004 04:24:24 +0000 (04:24 +0000)]
Fix Solaris thread build settings for non-gcc compilers.

20 years agomake
Tom Lane [Wed, 7 Jan 2004 23:03:06 +0000 (23:03 +0000)]
make

20 years agoMake some improvements in the intelligence of the partial-index
Tom Lane [Wed, 7 Jan 2004 22:02:48 +0000 (22:02 +0000)]
Make some improvements in the intelligence of the partial-index
predicate tester.  It can now deal with commuted clauses (for
instance, 4 < x implies x > 3), subclauses more complicated than
a simple Var (for example, upper(x) = 't' implies upper(x) > 'a'),
and <> operators (for example, x < 3 implies x <> 4).  Still
only understands operators associated with btree opclasses, though.
Inspired by example from Martin Hampl.

20 years agoTranslation updates
Dennis Bjorklund [Wed, 7 Jan 2004 21:51:13 +0000 (21:51 +0000)]
Translation updates

20 years agoAdd missing inclusion of nodes/value.h.
Tom Lane [Wed, 7 Jan 2004 21:12:56 +0000 (21:12 +0000)]
Add missing inclusion of nodes/value.h.

20 years agoMore janitorial work: remove the explicit casting of NULL literals to a
Neil Conway [Wed, 7 Jan 2004 18:56:30 +0000 (18:56 +0000)]
More janitorial work: remove the explicit casting of NULL literals to a
pointer type when it is not necessary to do so.

For future reference, casting NULL to a pointer type is only necessary
when (a) invoking a function AND either (b) the function has no prototype
OR (c) the function is a varargs function.

20 years agoCleanup: move the 'Value' node into a separate file, rather than putting
Neil Conway [Wed, 7 Jan 2004 18:43:36 +0000 (18:43 +0000)]
Cleanup: move the 'Value' node into a separate file, rather than putting
it in the same file as the 'List' node.

20 years agoNot done, per Tom:
Bruce Momjian [Wed, 7 Jan 2004 17:54:19 +0000 (17:54 +0000)]
Not done, per Tom:

>  o Allow parameters to be specified by name and type during definition

20 years agoSupport function parameter names in plpgsql. This is the last of
Tom Lane [Wed, 7 Jan 2004 06:20:02 +0000 (06:20 +0000)]
Support function parameter names in plpgsql.  This is the last of
Dennis Bjorklund's original patch for function parameter names, but
there's still plenty left to do (documentation for instance...)

20 years agoCompleted, I am pretty sure:
Bruce Momjian [Wed, 7 Jan 2004 04:32:56 +0000 (04:32 +0000)]
Completed, I am pretty sure:

>  o -Allow parameters to be specified by name and type during definition

20 years agopg_dump support for function parameter names.
Tom Lane [Wed, 7 Jan 2004 00:44:21 +0000 (00:44 +0000)]
pg_dump support for function parameter names.

20 years agoApply the core parts of Dennis Bjorklund's patch to allow function
Tom Lane [Tue, 6 Jan 2004 23:55:19 +0000 (23:55 +0000)]
Apply the core parts of Dennis Bjorklund's patch to allow function
parameters to be declared with names.  pg_proc has a column to store
names, and CREATE FUNCTION can insert data into it, but that's all as
yet.  I need to do more work on the pg_dump and plpgsql portions of the
patch before committing those, but I thought I'd get the bulky changes
in before the tree drifts under me.
initdb forced due to pg_proc change.

20 years agoFinal rearrangement of main postgresql child process (ie.
Bruce Momjian [Tue, 6 Jan 2004 23:15:22 +0000 (23:15 +0000)]
Final rearrangement of main postgresql child process (ie.
BackendFork/SSDataBase/pgstat) startup, to allow fork/exec calls to
closely mimic (the soon to be provided) Win32 CreateProcess equivalent
calls.

Claudio Natoli

20 years agoSuppress compiler warning (xlog_outrec is unused if not WAL_DEBUG).
Tom Lane [Tue, 6 Jan 2004 22:22:37 +0000 (22:22 +0000)]
Suppress compiler warning (xlog_outrec is unused if not WAL_DEBUG).

20 years agoFavor Freenode rather than EFnet in IRC listing. Freenode is more popular.
Bruce Momjian [Tue, 6 Jan 2004 21:44:07 +0000 (21:44 +0000)]
Favor Freenode rather than EFnet in IRC listing.  Freenode is more popular.

20 years agoDone:
Bruce Momjian [Tue, 6 Jan 2004 20:01:13 +0000 (20:01 +0000)]
Done:

> * -Have psql \dn show only visible temp schemas using current_schemas()

20 years agoCode cleanup, mostly in the smgr:
Neil Conway [Tue, 6 Jan 2004 18:07:32 +0000 (18:07 +0000)]
Code cleanup, mostly in the smgr:

     - Update comment in IsReservedName() to the present day

     - Improve some variable & function names in commands/vacuum.c. I
       was planning to rewrite this to avoid lappend(), but since I
       still intend to do the list rewrite, there's no need for that.

     - Update some smgr comments which seemed to imply that we still
       forced all dirty pages to disk at commit-time.

     - Replace some #ifdef DIAGNOSTIC code with assertions.

     - Make the distinction between OS-level file descriptors and
       virtual file descriptors a little clearer in a few comments

     - Other minor comment improvements in the smgr code

20 years agoRemove unused 'noversion' command-line option processing from the
Neil Conway [Tue, 6 Jan 2004 17:36:31 +0000 (17:36 +0000)]
Remove unused 'noversion' command-line option processing from the
backend.

20 years agoMake the 'wal_debug' GUC variable a boolean (rather than an integer), and
Neil Conway [Tue, 6 Jan 2004 17:26:23 +0000 (17:26 +0000)]
Make the 'wal_debug' GUC variable a boolean (rather than an integer), and
hide it behind #ifdef WAL_DEBUG blocks.

20 years agoInstead of rechecking lossy index operators by putting them into the
Tom Lane [Tue, 6 Jan 2004 04:31:01 +0000 (04:31 +0000)]
Instead of rechecking lossy index operators by putting them into the
regular qpqual ('filter condition'), add special-purpose code to
nodeIndexscan.c to recheck them.  This ends being almost no net addition
of code, because the removal of planner code balances out the extra
executor code, but it is significantly more efficient when a lossy
operator is involved in an OR indexscan.  The old implementation had
to recheck the entire indexqual in such cases.

20 years agoAdjust indexscan planning logic to keep RestrictInfo nodes associated
Tom Lane [Mon, 5 Jan 2004 23:39:54 +0000 (23:39 +0000)]
Adjust indexscan planning logic to keep RestrictInfo nodes associated
with index qual clauses in the Path representation.  This saves a little
work during createplan and (probably more importantly) allows reuse of
cached selectivity estimates during indexscan planning.  Also fix latent
bug: wrong plan would have been generated for a 'special operator' used
in a nestloop-inner-indexscan join qual, because the special operator
would not have gotten into the list of quals to recheck.  This bug is
only latent because at present the special-operator code could never
trigger on a join qual, but sooner or later someone will want to do it.

20 years agoTrivial refactoring: move analysis of ViewStmt into its own function for
Neil Conway [Mon, 5 Jan 2004 20:58:58 +0000 (20:58 +0000)]
Trivial refactoring: move analysis of ViewStmt into its own function for
readability and for the sake of consistency with the rest of analyze.c

20 years agoRemove an old and out-of-date SGML comment from this file: we already
Neil Conway [Mon, 5 Jan 2004 20:39:23 +0000 (20:39 +0000)]
Remove an old and out-of-date SGML comment from this file: we already
have WAL, outer joins, and TOAST.

20 years agoFix (yet another) instance of incorrect markup for a GUC variable.
Neil Conway [Mon, 5 Jan 2004 20:37:51 +0000 (20:37 +0000)]
Fix (yet another) instance of incorrect markup for a GUC variable.

20 years agoFix three trivial typos in comments.
Neil Conway [Mon, 5 Jan 2004 20:36:04 +0000 (20:36 +0000)]
Fix three trivial typos in comments.

20 years agoTranslation updates
Peter Eisentraut [Mon, 5 Jan 2004 19:15:48 +0000 (19:15 +0000)]
Translation updates

20 years agoImprove UniquePath logic to detect the case where the input is already
Tom Lane [Mon, 5 Jan 2004 18:04:39 +0000 (18:04 +0000)]
Improve UniquePath logic to detect the case where the input is already
known unique (eg, it is a SELECT DISTINCT ... subquery), and not do a
redundant unique-ification step.

20 years agoDept. of second thoughts: clause_selectivity shouldn't try to cache its
Tom Lane [Mon, 5 Jan 2004 16:44:40 +0000 (16:44 +0000)]
Dept. of second thoughts: clause_selectivity shouldn't try to cache its
result for jointypes associated with IN processing.

20 years agoAdd the ability to extract OR indexscan conditions from OR-of-AND
Tom Lane [Mon, 5 Jan 2004 05:07:36 +0000 (05:07 +0000)]
Add the ability to extract OR indexscan conditions from OR-of-AND
join conditions in which each OR subclause includes a constraint on
the same relation.  This implements the other useful side-effect of
conversion to CNF format, without its unpleasant side-effects.  As
per pghackers discussion of a few weeks ago.

20 years agoUpdate for 7.4.1 as newest version.
Bruce Momjian [Mon, 5 Jan 2004 00:54:24 +0000 (00:54 +0000)]
Update for 7.4.1 as newest version.

20 years agoCorrect gettext URL.
Peter Eisentraut [Sun, 4 Jan 2004 16:43:52 +0000 (16:43 +0000)]
Correct gettext URL.

20 years agoFixed bug in GRANT OPTION FOR parsing.
Michael Meskes [Sun, 4 Jan 2004 14:49:06 +0000 (14:49 +0000)]
Fixed bug in GRANT OPTION FOR parsing.

20 years agoThere's no longer any good reason for genbki.sh and Gen_fmgrtab.sh to
Tom Lane [Sun, 4 Jan 2004 05:57:21 +0000 (05:57 +0000)]
There's no longer any good reason for genbki.sh and Gen_fmgrtab.sh to
run the data through cpp, and we know of at least one platform where
unusual cpp behavior breaks the process.  So remove the cpp step,
and make consequent simplifications.

20 years agoFix portability bugs: char values passed to <ctype.h> functions must
Tom Lane [Sun, 4 Jan 2004 04:17:01 +0000 (04:17 +0000)]
Fix portability bugs: char values passed to <ctype.h> functions must
be cast to unsigned char.  We have learned this the hard way before.

20 years agoFix discrepancy in prototypes for HPUX pg_dlerror.
Tom Lane [Sun, 4 Jan 2004 04:06:37 +0000 (04:06 +0000)]
Fix discrepancy in prototypes for HPUX pg_dlerror.

20 years agoFix ReadOffset() to work correctly when off_t is wider than int.
Tom Lane [Sun, 4 Jan 2004 04:02:15 +0000 (04:02 +0000)]
Fix ReadOffset() to work correctly when off_t is wider than int.

20 years agoMerge restrictlist_selectivity into clauselist_selectivity by
Tom Lane [Sun, 4 Jan 2004 03:51:52 +0000 (03:51 +0000)]
Merge restrictlist_selectivity into clauselist_selectivity by
teaching the latter to accept either RestrictInfo nodes or bare
clause expressions; and cache the selectivity result in the RestrictInfo
node when possible.  This extends the caching behavior of approx_selectivity
to many more contexts, and should reduce duplicate selectivity
calculations.

20 years agoRemove trailing newline from file.
Bruce Momjian [Sun, 4 Jan 2004 00:50:11 +0000 (00:50 +0000)]
Remove trailing newline from file.

20 years agoRemove trailing semicolons from macro initializations in plpython.
Bruce Momjian [Sun, 4 Jan 2004 00:14:17 +0000 (00:14 +0000)]
Remove trailing semicolons from macro initializations in plpython.

Problem report on True64 Unix by Nikola Milutinovic.

20 years agoMention grabbing typedefs from pgsql/lib too.
Bruce Momjian [Sun, 4 Jan 2004 00:11:29 +0000 (00:11 +0000)]
Mention grabbing typedefs from pgsql/lib too.

20 years agoRewrite OR indexscan processing to be more flexible. We can now for the
Tom Lane [Sun, 4 Jan 2004 00:07:32 +0000 (00:07 +0000)]
Rewrite OR indexscan processing to be more flexible.  We can now for the
first time generate an OR indexscan for a two-column index when the WHERE
condition is like 'col1 = foo AND (col2 = bar OR col2 = baz)' --- before,
the OR had to be on the first column of the index or we'd not notice the
possibility of using it.  Some progress towards extracting OR indexscans
from subclauses of an OR that references multiple relations, too, although
this code is #ifdef'd out because it needs more work.

20 years agoMust test for __hppa__ as well as __hppa to make linux-hppa happy.
Tom Lane [Sat, 3 Jan 2004 05:47:44 +0000 (05:47 +0000)]
Must test for __hppa__ as well as __hppa to make linux-hppa happy.

20 years agoDo an explicit fflush after writing a progress message with puts.
Tom Lane [Thu, 1 Jan 2004 19:27:15 +0000 (19:27 +0000)]
Do an explicit fflush after writing a progress message with puts.
This ensures stdout is kept in sync with messages on stderr.
Per report from Olaf Ferger.

20 years agoMake Solaris non-gcc thread flag "-mt" a compile flag, not a link flag.
Bruce Momjian [Wed, 31 Dec 2003 16:14:44 +0000 (16:14 +0000)]
Make Solaris non-gcc thread flag "-mt" a compile flag, not a link flag.

20 years agoAdjust the definition of RestrictInfo's left_relids and right_relids
Tom Lane [Tue, 30 Dec 2003 23:53:15 +0000 (23:53 +0000)]
Adjust the definition of RestrictInfo's left_relids and right_relids
fields: now they are valid whenever the clause is a binary opclause,
not only when it is a potential join clause (there is a new boolean
field canjoin to signal the latter condition).  This lets us avoid
recomputing the relid sets over and over while examining indexes.
Still more work to do to make this as useful as it could be, because
there are places that could use the info but don't have access to the
RestrictInfo node.