Time for 7.2.1? - Mailing list pgsql-hackers

Seems like it's about time to release 7.2.1.  We have enough
post-release patches in there that a dot-release is clearly needed
(particularly for the pg_dump-related issues).  And it seems like
not too much new stuff is coming in.  Does anyone have any issues
outstanding that need to be dealt with before 7.2.1?
        regards, tom lane


Post-release patches in the 7.2 branch:

2002-03-05 01:10  momjian
* contrib/tsearch/dict/porter_english.dct (REL7_2_STABLE): Please,apply attached patch for contrib/tsearch to 7.2.1 and
currentCVS.It  fix english stemmer's problem with ending words like'technology'.We have found one more bug in english
stemmer.The bug is with'irregular'  english words like 'skies' -> 'sky'. Please, applyattached cumulative patch to
7.2.1and current CVS insteadprevious one.Thank to Thomas T. Thai <tom@minnesota.com> for hard testing. Thiskind of bug
hassignificance only for dump/reload database andviewing, but searching/indexing works right.Teodor Sigaev
 

2002-03-05 00:13  tgl
* src/backend/optimizer/prep/prepunion.c (REL7_2_STABLE): Previouspatch to mark UNION outputs with common typmod (if
any)breaksthree-or-more-way UNIONs, as per example from Josh Berkus.  Causeis a fragile assumption that one tlist's
entrieswill exactly matchanother.  Restructure code to make that assumption a little lessfragile.
 

2002-03-04 22:45  ishii
* src/: backend/utils/adt/timestamp.c,test/regress/expected/timestamp.out,test/regress/expected/timestamptz.out
(REL7_2_STABLE):A backportpatch:Fix bug in extract/date_part for milliseconds/miscroseconds andtimestamp/timestamptz
combo.Now extract/date_part returnsseconds*1000 or 1000000 + fraction part as the manual stats. regression test are
alsofixed.See the thread in pgsql-hackers:Subject: Re: [HACKERS] timestamp_part() bug? Date: Sat, 02 Mar 200211:29:53
+0900

2002-03-04 12:47  tgl
* doc/FAQ_Solaris (REL7_2_STABLE): Update FAQ_Solaris with infoabout gcc 2.95.1 problems and how to work around 64-bit
vsnprintfbug.

2002-02-27 18:17  tgl
* src/backend/tcop/postgres.c (REL7_2_STABLE): Back-patch fix forerrors reported at transaction end.

2002-02-26 20:47  ishii
* src/backend/commands/copy.c (REL7_2_STABLE): Back-patch fix forfollowings:Fix bug in COPY FROM when DELIMITER is not
inASCII range.  Seepgsql-bugs/pgsql-hackers discussion "COPY FROM is not 8bit clean"around 2002/02/26 for more details
--Tatsuo Ishii
 

2002-02-26 18:48  tgl
* src/: backend/commands/command.c, backend/commands/explain.c,backend/executor/functions.c,
backend/executor/spi.c,backend/nodes/copyfuncs.c,backend/nodes/equalfuncs.c,backend/nodes/readfuncs.c,
backend/parser/analyze.c,backend/tcop/dest.c,backend/tcop/postgres.c,backend/tcop/pquery.c,
backend/tcop/utility.c,include/commands/command.h,include/nodes/parsenodes.h,include/tcop/dest.h,
include/tcop/pquery.h,include/tcop/utility.h(REL7_2_STABLE): Back-patch fix for command completion reporthandling.
Thisis primarily needed so that INSERTing a row stillreports the row's OID even when there are ON INSERT rules
firingadditionalqueries.
 

2002-02-25 16:37  tgl
* src/bin/psql/command.c (REL7_2_STABLE): Tweak psql's \connectcommand to not downcase unquoted database and user
names. This isa temporary measure to allow backwards compatibility with 7.2 andearlier pg_dump.  7.2.1 and later
pg_dumpwill double-quote mixedcase names in \connect.  Once we feel that older dumps are not aproblem anymore, we can
revertthis change and treat \connectarguments as normal SQL identifiers.
 

2002-02-25 15:07  momjian
* src/backend/libpq/auth.c (REL7_2_STABLE): Fix for PAM errormessage display:> and that the right fix is to make each
ofthe subsequent calls bein> this same pattern, not to try to emulate their nonsensical style.Dominic J. Eidson
 

2002-02-25 11:22  thomas
* src/backend/utils/adt/datetime.c (REL7_2_STABLE): Add a largenumber of time zones to the lookup table.  Fix a
fewapparently-wrongTZ vs DTZ declarations.  Same patch as added toHEAD.
 

2002-02-22 10:40  momjian
* src/include/libpq/pqsignal.h (REL7_2_STABLE): We had a problemwith to compile pgsql-7.2 under SW-8.0.  In the mailing
listsIfound no informations.    See note for further informations.Add missing AuthBlockSig.regards Heiko
 

2002-02-22 08:02  momjian
* doc/: FAQ_russian, src/FAQ/FAQ_russian.html (REL7_2_STABLE):BACKPATCH:Add Russian FAQ to 7.2.1.  Why not?

2002-02-22 01:08  momjian
* contrib/btree_gist/btree_gist.c (REL7_2_STABLE): BACKPATCH:Please, apply attached patch of contrib/btree_gist to
7.2.1andcurrent cvs.  The patch fixes memory leak during creation GiSTindex on timestamp column.Thank you.-- Teodor
Sigaevteodor@stack.net
 

2002-02-19 17:19  tgl
* src/backend/utils/adt/cash.c (REL7_2_STABLE): Avoid failures incash_out and cash_words for INT_MIN.  Also, 'fourty'
->'forty'.
 

2002-02-18 11:04  tgl
* src/backend/commands/analyze.c (REL7_2_STABLE): Replacenumber-of-distinct-values estimator equation, per recent
pghackersdiscussion.

2002-02-17 23:12  ishii
* src/bin/pgaccess/lib/tables.tcl (REL7_2_STABLE): Fixkanji-coversion key binding. This has been broken since 7.1
PerYoshinoriAriie's report.
 

2002-02-17 08:29  momjian
* doc/src/sgml/ref/alter_table.sgml: Fix SGML typo in previouspatch.

2002-02-17 06:50  momjian
* doc/src/sgml/ref/alter_table.sgml: I think it's important thatit's actually documented that they can add primary keys
afterthefact!Also, we need to add regression tests for alter table / add primarykey and alter table / drop constraint.
 These shouldn't be addeduntil 7.3 tho methinks...Chris
 

2002-02-16 18:45  momjian
* doc/src/sgml/ref/alter_table.sgml: Clarify params to ALTER TABLEto clearly show single parameters.e.g. table
contraintdefinition -> table_constraint_definition.
 

2002-02-15 12:46  petere
* src/interfaces/ecpg/preproc/pgc.l: Remove warning about automaticinclusion of sqlca.

2002-02-14 10:24  tgl
* src/: backend/commands/command.c, backend/executor/spi.c,backend/utils/mmgr/portalmem.c, include/utils/portal.h:
Ensurethata cursor is scanned under the same scanCommandId it was originallycreated with, so that the set of visible
tuplesdoes not change asa result of other activity.  This essentially makes PG cursorsINSENSITIVE per the SQL92
definition. See bug report of 13-Feb-02.
 

2002-02-13 14:32  tgl
* doc/src/sgml/ref/createuser.sgml: Point out that --adduseractually makes the new user a superuser.  This was
mentionedon theman page for the underlying CREATE USER command, but it should beexplained here too.
 

2002-02-12 18:39  tgl
* src/backend/port/dynloader/: README.dlfcn.aix, aix.h, bsdi.h,dgux.h, freebsd.h, irix5.h, linux.h, netbsd.h,
openbsd.h,osf.h,sco.h, solaris.h, sunos4.h, svr4.h, univel.h, unixware.h, win.h:Use RTLD_NOW, not RTLD_LAZY, as binding
modefor dlopen() on allplatforms.  This restores the Linux behavior to what it was in PG7.0 and 7.1, and causes other
platformsto agree.  (Otherwell-tested platforms like HPUX were doing it this way already.) Per pghackers discussion
overthe past month or so.
 

2002-02-12 17:35  tgl
* doc/FAQ_Solaris: Add warning not to use /usr/ucb/cc on Solaris.

2002-02-12 17:25  momjian
* doc/src/sgml/advanced.sgml: Fix tutorial for references problem,from rainer.tammer@spg.schulergroup.com

2002-02-12 16:25  tgl
* doc/src/sgml/ref/copy.sgml, src/backend/commands/copy.c: ModifyCOPY TO to emit carriage returns and newlines as
backslashescapes(backslash-r, backslash-n) for protection againstnewline-conversion munging.  In future we will also
tweakCOPYFROM, but this part of the change should be backwards-compatible. Per pghackers discussion.  Also, update COPY
referencepage todescribe the backslash conversions more completely and accurately. 

2002-02-11 18:25  momjian
* doc/src/sgml/wal.sgml: Update wal files computationdocumentation.

2002-02-11 17:41  tgl
* src/backend/access/gist/gist.c: Tweak GiST code to work correctlyon machines where 8-byte alignment of pointers is
required. Patchfrom Teodor Sigaev per pghackers discussion.  It's an ugly klugebut avoids forcing initdb; we'll put a
betterfix into 7.3 orlater.
 

2002-02-11 16:38  petere
* src/backend/port/dynloader/freebsd.h: Fix for old FreeBSDversions that don't have RTLD_GLOBAL

2002-02-11 15:10  tgl
* src/backend/executor/: nodeIndexscan.c, nodeTidscan.c: Repairproblems with EvalPlanQual where target table is scanned
asinnerindexscan (ie, one with runtime keys).    ExecIndexReScan mustcompute or recompute runtime keys even if we are
rescanningin theEPQ case.  TidScan seems to have comparable problems.  Per bugnoted by Barry Lind 11-Feb-02.
 

2002-02-11 10:19  momjian
* contrib/pg_upgrade/pg_upgrade: Fix flag handling of pg_upgrade.

2002-02-10 19:18  tgl
* src/bin/pg_dump/: common.c, pg_backup.h, pg_backup_archiver.c,pg_dump.c, pg_dump.h, pg_dumpall.sh: Be more wary about
mixed-casedatabasenames and user names.    Get the CREATE DATABASE commandright in pg_dump -C case.
 

2002-02-10 19:14  tgl
* doc/src/sgml/ref/: pg_dump.sgml, pg_restore.sgml: pg_dump andpg_restore man pages need to mention that one should
restoreinto avirgin database, ie, one created from template0, if there are anysite-local additions in template1.
 

2002-02-10 17:56  tgl
* src/backend/storage/file/fd.c: Don't Assert() that fsync() andclose() never fail; I have seen this crash on Solaris
whenoverdisk quota.  Instead, report such failures via elog(DEBUG).
 

2002-02-08 11:30  momjian
* src/backend/utils/init/findbe.c: Move sys/types.h to top, forhiroyuki hanai/ FreeBSD.

2002-02-08 09:47  momjian
* contrib/mysql/my2pg.pl: Upgrade my2pg version 1.23.

2002-02-07 17:20  tgl
* src/backend/postmaster/pgstat.c: pgstat's truncation of querystring needs to be multibyte-aware.  Patch from
sugita@sra.co.jp.

2002-02-07 17:11  tgl
* contrib/: intarray/_int.c, tsearch/README.tsearch,tsearch/gistidx.c, tsearch/tsearch.sql.in: Repair some problems
inGIST-indexcontrib modules.  Patch from Teodor Sigaev<teodor@stack.net>.
 

2002-02-06 19:27  inoue
* src/backend/tcop/utility.c: Removed a check for REINDEX TABLE.

2002-02-06 15:29  petere
* doc/Makefile: Fix for parallel make

2002-02-06 12:27  tgl
* src/bin/pg_dump/: pg_backup_archiver.c, pg_dump.c: Fix failure toreconnect as sequence's owner before issuing
setval().