Skip to content

Wöchentlicher PostgreSQL Newsletter - 15. Februar 2009

Der Originalartikel befindet sich unter:

== Wöchentlicher PostgreSQL Newsletter - 15. Februar 2009 ==

Auf dem Prager PostgreSQL Developer Tag haben die Tschechischen und
Slovakischen Leute die Wiedervereinigung beschlossen, um eine
Tschechoslowakische PostgreSQL User Gruppe zu formen. Erwarte baldige

Termin vormerken: pgDay San Jose, am Sonntag dem 19. Juli unmittelbar
vor der OSCON. CfP und mehr Informationen bald.

== PostgreSQL Produkt Neuigkeiten ==

Benetl 2.5, ein ETL Programm für Dateien die PostgreSQL nutzen, ist

pgpool-II 2.2 RC1, ein Connection Pooler und mehr, ist erschienen.

== PostgreSQL 8.4 Feature der Woche ==

Paralleles Wiederherstellen. Andrew Dunstan hat pg_restore
parallelisiert, so das du alle deine Prozessoren ausnutzen (für ein
Upgrade) der Datenbank nutzen kannst. In Tests hat das Tool den Load um
50% auf 4 Kernen und um 75% auf 8 Kernen reduziert. Und noch besser:
das neue pg_restore funktioniert mit 8.3 Datenbanken.

== PostgreSQL Tipp der Woche ==

Kein Index auf Spalten mit geringer Kardinalität, ausser sie sind Teil
eines Index über mehrere Spalten.

== PostgreSQL Jobs im Februar ==

== PostgreSQL Lokal ==

Ed Borasky wird über Datenvisualisierung vor der PDXPUG am 19. Februar
2009 von 19-21 Uhr in Portland sprechen.

PostgreSQL wird auf der SCALE vom 20. bis 22. Februar in Los Angeles
vertreten sein.

Andreas 'ads' Scherbaum wird einen Vortrag auf dem Perl Workshop 2009
am 25. Februar in Frankfurt am Main, Deutschland, halten.

Die Deutsche PostgreSQL User Gruppe wird einen Stand, 2 Vorträge und
einen Workshop auf den Chemnitzer Linuxtagen am 14. und 15. März 2009
haben. Mehr Informationen auf Deutsch hier:

PostgreSQL Conference U.S. wird einen PgDay auf dem Linuxfest Nordwest
am 25. und 26. April veranstalten. Der Call for Papers ist unter:

PGCon 2009 wird vom 21. bis 22. Mai 2009 in Ottawa an der Universität
von Ottawa stattfinden. Dem gehen zwei Tage mit Tutorials vom 19. bis
20. Mai 2009 voraus.

PGCon Brasilien wird am 23.-24. Oktober 2009 auf dem Unicamp in
Campinas, Sao Paulo, stattfinden.

== PostgreSQL in den News ==

Planet PostgreSQL:

Dieser wöchentliche PostgreSQL Newsletter wurde erstellt von David
Fetter und Josh Berkus.

Sende Neuigkeiten und Ankündigungen bis Sonntag, 15 Uhr Pazifischer
Zeit. Bitte sende englische Beiträge an, deutsche an, italienische an

== Angewandte Patches ==

Bruce Momjian committed:

- In pgsql/doc/src/sgml/ref/psql-ref.sgml, fix spelling error in docs.
  Erik Rijkers

- In pgsql/doc/src/sgml/ref/pg_dump.sgml, update wording of how to
  prevent pg_dump from affecting statistics collection.

- In pgsql/doc/src/sgml/libpq.sgml, clarify PQinitSSL() documentation
  to mention what the argument controls.

- In pgsql/doc/src/sgml/ref/create_table.sgml, remove tabs from SGML.

Alvaro Herrera committed:

- Update autovacuum to use reloptions instead of a system catalog, for
  per-table overrides of parameters.  This removes a whole class of
  problems related to misusing the catalog, and perhaps more
  importantly, gives us pg_dump support for the parameters.  Based on
  a patch by Euler Taveira de Oliveira, heavily reworked by me.

- In pgsql/src/bin/psql/describe.c, improve psql \d+ to show TOAST
  reloptions too.  Per complaint from ITAGAKI Takahiro.

Tom Lane committed:

- Adopt Bob Jenkins' improved hash function for hash_any().  This
  changes the contents of hash indexes (again), so bump catversion.
  Kenneth Marshall.

- Tweak configure to attempt to add -qnoansialias to CFLAGS whenever
  running on AIX with a non-gcc compiler.  The previous coding would
  do this only if CC was exactly "xlc"; which is a bad idea, as
  demonstrated by trouble report from Mihai Criveti.

- Change ALTER TABLE SET WITHOUT OIDS to rewrite the whole table to
  physically get rid of the OID column.  This eliminates the problem
  discovered by Heikki back in November that 8.4's suppression of
  "unnecessary" junk filtering in INSERT/SELECT could lead to an
  Assert failure, or storing of oids into a table that shouldn't have
  them if Asserts are off.  While that particular problem could have
  been solved in other ways, it seems likely to be just a forerunner
  of things to come if we continue to allow tables to contain rows
  that disagree with the pg_class.relhasoids setting.  It's better to
  make this operation slow than to sacrifice performance or risk bugs
  in more common code paths.  Also, add ALTER TABLE SET WITH OIDS to
  rewrite the table to add oids.  This was a bit more controversial,
  but in view of the very small amount of extra code needed given the
  current ALTER TABLE infrastructure, it seems best to eliminate the
  asymmetry in features.

- Fix UNLISTEN to fall out quickly if the current backend has never
  executed any LISTEN command.  This is more important than it used to
  be because DISCARD ALL invokes UNLISTEN.  Connection-pooled
  applications making heavy use of DISCARD ALL were seeing significant
  contention for pg_listener, as reported by Matteo Beccati.  It seems
  unlikely that clients using LISTEN would use pooled connections, so
  this simple tweak seems sufficient, especially since the pg_listener
  implementation is slated to go away soon anyway.  Back-patch to 8.3,
  where DISCARD ALL was introduced.

- In pgsql/src/backend/catalog/information_schema.sql, a couple of
  marginal performance hacks for the information_schema views: replace
  the old recursive-SQL-function implementation of _pg_keysequal()
  with use of the built-in array containment operators, and change
  table_constraints' UNION to UNION ALL.  Per discussion with Octavio
  Alvarez.  initdb not forced since this doesn't affect results, but
  you'd need to initdb or reload the information_schema to see the new

- In pgsql/doc/src/sgml/config.sgml, minor wordsmithing.

- Teach the planner to treat a partial unique index as proving a
  variable is unique for a particular query, if the index predicate is
  satisfied.  This requires a bit of reordering of operations so that
  we check the predicates before doing any selectivity estimates, but
  shouldn't really cause any noticeable slowdown.  Per a comment from
  Michal Politowski.

Peter Eisentraut committed:

- In pgsql/src/backend/utils/mb/Unicode/, add
  possibility to generate only some files, by passing command-line

- In pgsql/src/backend/utils/mb/wchar.c, remove the encoding *numbers*
  from the comments.  They are useless, and make maintenance harder.

- Support for KOI8U encoding.

- Only unset the locale environment when --no-locale is used (the way
  it was presumably designed, but didn't act).  This allows running
  the temp install tests in a non-C locale, thus exercising users'
  real environments better.  Document how to change locales for test

- Unset language-related locale settings in any case, otherwise psql
  will speak in tongues and mess up the regression test diffs.

- The Czech (cs_CZ) and Slovak (sk_SK) locales sort numbers after
  letters, instead of vice versa.  Update the regression test
  expectations to support that.  In the plpgsql test, adjust the test
  data so that this isn't an issue.  In the char and varchar tests,
  add new expected files.

Magnus Hagander committed:

- In pgsql/src/port/win32env.c, don't call SetEnvironmentVariable()
  when removing an environment variable, as this seems to crash on at
  least some versions of MingW. Our current usage of this function
  does not require it, so it should be ok to ignore.

- In pgsql/src/port/kill.c, loop calling CallNamedPipe() several times
  in case it fails, since it can be transient failures, causing kill()
  to not properly send signals.  Original patch from Steve Marshall,
  modified by me.

Andrew Dunstan committed:

- Do not use unsetenv.c on mingw.

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Heikki Linnakangas sent in another revision of the recovery
infrastructure patch.

Heikki Linnakangas sent in a change to the way subtransaction cache
works.  This is in aid of Hot Standby.

ITAGAKI Takahiro sent in a patch to reject toast.fillfactor as a
reloption, modify the relopt_kind to bit flags, and report the
relation type on "unrecognized parameter" errors.

Teodor Sigaev sent in another revision of the fast GIN insert patch.

Pavel Stehule sent in a patch to add in some parser hooks for
Oracle-compatible DECODE support.

Jeff Davis sent in a patch to advance xmin more aggressively in
long-running transactions.

KaiGai Kohei sent in two more revisions of the SE-PostgreSQL patch.

Alvaro Herrera sent in two revisions of a patch to adjust psql's
describe to do the right thing in when talking to different versions
of PostgreSQL (pre-8.4 vs. later) pursuant to the reloptions patch.

Fujii Masao sent in another revision of the synchronous replication

Teodor Sigaev sent in another revision of the B-Tree emulation for GIN

Andrew Chernow sent in two revisions of a patch which adds a
PQInitCrypto(int) function to libpq so that it's possible to use a
crypto library separately from SSL.


No Trackbacks


Display comments as Linear | Threaded

No comments

Add Comment

Enclosing asterisks marks text as bold (*word*), underscore are made via _word_.
E-Mail addresses will not be displayed and will only be used for E-Mail notifications.
To leave a comment you must approve it via e-mail, which will be sent to your address after submission.
Form options