Skip to content

Wöchentlicher PostgreSQL Newsletter - 14. Juni 2009

Der Originalartikel befindet sich unter:

== Wöchentlicher PostgreSQL Newsletter - 14. Juni 2009 ==

PostgreSQL 8.4 RC1 wird bald erscheinen. Testen!

Neue Umfrage: Was ist deine höchste Priorität für die PostgreSQL

Das TikiWiki Projekt braucht PostgreSQL Hilfe! TikiWiki braucht einen
neuen Maintainer für die PostgreSQL Kompatibilität, andernfalls wird der
Support abgeschaltet. Wenn du TikiWiki nutzt und helfen kannst,
kontaktiere bitte marclaporte AT tikiwiki DOT org.

== PostgreSQL Produkt Neuigkeiten ==

Mogwai ER-DesignerNG 1.7-SNAPSHOT, ein Java-basiertes ERD Werkzeug, ist

Slony1-ctl 1.1.4, ein Set von Skripts, um Slony zu verwalten, ist

== PostgreSQL 8.4 Feature der Woche ==

Das zusätzliche (Contrib) Modul pg_stat_statements, welches serverweit
die Statistiken zu ausgeführten Anfragen protokolliert.
Kapazitätsplaner werden sich freuen!

== PostgreSQL Tipp der Woche==

Die referenzierte ("Child") Seite einer Foreign Key Spalte sollte
generell indiziert sein, außer die referenzierten Werte haben eine sehr
geringe Kardinalität. PostgreSQL erledigt dies nicht automatisch.

== PostgreSQL Jobs im Juni ==

== PostgreSQL Lokal ==

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

Der CfP für PGCon Brasilien ist eröffnet!

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

PGDay.EU 2009 wird an der Telecom ParisTech Universität in Paris,
Frankreich, am 6. und 7. Nivember 2009 stattfinden.

Die 10. jährliche JPUG Konferenz hat den Aufruf für Vorträge gestartet.
Die Konferenz findet am 20-21. November 2009 in Tokio, Japan, statt.

== 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 ==

Tom Lane committed:

- In pgsql/doc/src/sgml/release-8.4.sgml, be a bit more verbose about
  the effects of string literal processing changes in plpgsql.  Per
  bug #4843.

- Fix contrib/pageinspect to not create an ABI breakage between 8.3
  and 8.4.  The original implementation of the 3-argument form of
  get_raw_page() risked core dumps if the 8.3 SQL function definition
  was mistakenly used with the 8.4 module, which is entirely likely
  after a dump-and-reload upgrade.  To protect 8.4 beta testers
  against upgrade problems, add a check on PG_NARGS.  In passing, fix
  missed additions to the uninstall script, and polish the docs a

- In pgsql/src/backend/utils/adt/ruleutils.c, fix failure to
  double-quote function argument names when needed, in
  pg_get_function_arguments() and related functions.  Per report from
  Andreas Nolte.

- Restore dblink_current_query() to being a C-language function, so as
  to not create an ABI break between 8.3 and 8.4.  It is still just a
  wrapper around the built-in current_query() function, but at a
  different implementation level.  Per my proposal.  Note: this change
  doesn't break 8.4beta installations, since their SQL-language
  definition of the function still works fine.

- In pgsql/src/backend/utils/adt/array_userfuncs.c, switch order of
  tests to avoid possible Assert failure for
  "array_agg_finalfn(null)".  We should modify pg_proc entries to
  prevent this query from being accepted, but let's just make the
  function itself secure too.  Per my note of today.

- In pgsql/src/include/catalog/pg_proc.h, mark internal_in as not
  strict, so that the construct "null::internal" will throw an error,
  rather than possibly allowing someone to synthesize a manual call to
  an internal-accepting function.  As of CVS HEAD and existing
  releases, all such functions are either STRICT or careful about null
  inputs, so there is no current security issue here.  But it seems
  like a good idea to lock this down to protect against future
  mistakes.  In passing, similarly lock down trigger_in,
  language_handler_in, opaque_in, and shell_in.  These are not
  believed to present any security risk, but there's still no good
  reason to allow nulls of these types to be created.  I left the
  polymorphic pseudotypes (anyelement etc) alone, since a null of one
  of those types doesn't seem to be a problem --- the worst you can
  say about it is that it doesn't have an underlying non-polymorphic
  type.  If we were to make this change during normal development,
  we'd just automatically bump catversion for a pg_proc.h change.  But
  since this doesn't create a compatibility risk and isn't believed to
  be fixing a live bug, it seems better not to force a catversion bump
  in late beta.

- In pgsql/src/backend/utils/adt/xml.c, ensure xmlFree(NULL) is a
  no-op instead of a core dump.  Per report from Sergey Burladyan,
  there are at least some dank corners of libxml2 that assume this
  behavior, even though their published documentation suggests they
  shouldn't.  This is only really a live problem in 8.3, but the code
  is still there for possible debugging use in HEAD, so patch both

  input more consistent with other cases, by having an unlabeled
  integer field be treated as a number of minutes or seconds
  respectively.  These cases are outside the spec (which insists on
  full "dd hh:mm" or "dd hh:mm:ss" input respectively), so it's not
  much help to us in deciding what to do.  But with this change, it's
  uniformly the case that an unlabeled integer will be considered as
  being a number of the interval's rightmost field.  The change also
  takes us back to the 8.3 behavior of throwing error for certain
  ambiguous inputs such as INTERVAL '1 2' DAY TO MINUTE.  Per recent

- In pgsql/src/backend/utils/adt/cash.c, fix cash_in() to behave
  properly in locales where frac_digits is zero, eg Japan.  Report and
  fix by Itagaki Takahiro.  Also fix CASHDEBUG printout format for
  branches with 64-bit money type, and some minor comment cleanup.
  Back-patch to 7.4, because it's broken all the way back.

- In pgsql/src/backend/access/heap/heapam.c, keep rs_startblock the
  same during heap_rescan, so that a rescan of a SeqScan node starts
  from the same place as the first scan did.  This avoids surprising
  behavior of scrollable and WITH HOLD cursors, as seen in Mark
  Kirkwood's bug report of yesterday.  It's not entirely clear whether
  a rescan should be forced to drop out of the syncscan mode, but for
  the moment I left the code behaving the same on that point.  Any
  change there would only be a performance and not a correctness
  issue, anyway.  Back-patch to 8.3, since the unstable behavior was
  created by the syncscan patch.

- In pgsql/doc/src/sgml/ref/declare.sgml, add a warning about possible
  strange behavior of volatile functions in cursors.  This has always
  been the case, but given the lack of user complaints about it, I'm
  not going to bother back-patching this.

- In pgsql/doc/src/sgml/datatype.sgml, add warning that xpath()
  doesn't work with non-UTF8 data.

- Support Kerberos on platforms where libcom_err depends on OpenSSL,
  as reportedly is true on OpenBSD.  Also support OpenBSD's spelling
  of -Wl,--as-needed.  Per Simon Bertrang.

- Properly mark pg_freespace() function as strict.  Also update
  uninstall script to match reality.

- Fix ill-advised whitespace prettification, per buildfarm.

- In pgsql/src/backend/foreign/foreign.c, somebody seems to have
  thought they could get away without checking for
  rsinfo->expectedDesc == NULL in deflist_to_tuplestore(), but that
  doesn't look very safe to me.  Noted in passing while studying
  problem report from Greg Davidson.

- Fix things so that you can still do "select foo()" where foo is a
  SQL function returning setof record.  This used to work, more or
  less accidentally, but I had broken it while extending the code to
  allow materialize-mode functions to be called in select lists.  Add
  a regression test case so it doesn't get broken again.  Per gripe
  from Greg Davidson.

- Update time zone data files to tzdata release 2009i: DST law changes
  in Bangladesh, Egypt, Jordan, Pakistan.

- Mark contrib's GiST and GIN opclass support functions as STRICT, for
  safety.  (Note: GiST penalty functions could possibly be non-strict,
  but none are at present.)

- Remove our inadequate kluge that tried to get AIX's various broken
  versions of getaddrinfo() to work.  Instead, recommend updating the
  OS to get a working version of getaddrinfo.  Per recent discussions.

- Revisit AlterTableCreateToastTable's API once again, hoping to make
  it what pg_migrator actually needs and not just a partial solution.
  We have to be able to specify the OID that the new toast table
  should be created with.

- In pgsql/doc/src/sgml/release-8.4.sgml, update release notes to

- In pgsql/doc/src/sgml/installation.sgml, add info about minimum
  recommended AIX fix levels, per Laurenz Albe.

- Fix several places where a function was declared static and then
  defined without static.  Per testing with a compiler that complains
  about this.

- In pgsql/doc/src/sgml/gist.sgml, improve documentation about GiST
  opclass support functions.  Dimitri Fontaine.

- In pgsql/src/backend/parser/parse_oper.c, fix
  get_sort_group_operators() so that it doesn't think arrays can be
  grouped via hashing.  Eventually we ought to make that possible, but
  it won't happen for 8.4.  Per yesterday's report from Robert Haas.

- Change test tables in copy2 regression test to be temporary tables.
  This prevents autovacuum from reclaiming free space in them and
  causing the test's output row order to change, which is causing
  intermittent bogus failure reports in the buildfarm.  Backpatch to
  8.3.  The issue exists further back, but since autovacuum was not on
  by default before 8.3, it's not a problem for buildfarm testing.

Peter Eisentraut committed:

- In pgsql/src/backend/utils/adt/xml.c, fix map_sql_table_to_xmlschema()
  with dropped attributes.  Also backpatched to 8.3.

- Fix xmlattribute escaping XML special characters twice (bug #4822).
  ITAGAKI Takahiro.

- Correct/improve the datetime_precision field in the information
  schema.  In particular, always show 0 for the date type instead of
  null, and show 6 (the default) for time, timestamp, and interval
  without a declared precision.  This is now in fuller conformance
  with the SQL standard.  Also clarify the documentation about this.
  Discovered and analyzed by Konstantin Izmailov and Tom Lane

- In pgsql/src/backend/access/gist/gist.c, improve capitalization and
  punctuation in recently added GiST message.

- In pgsql/src/bin/psql/describe.c, clarify error message.

- In pgsql/src/pl/plpgsql/src/scan.l, clarify to the translator that
  plpgsql_yyerror() deals with the translation of "syntax error", not
  the literal string.  This same change was made in the backend a
  while ago; but it applies to plpgsql as well.

- In ecpg, arrange that the string "syntax error" generated by bison
  is translated.

- Translation updates.

- In pgsql/src/bin/psql/describe.c, tweak the display of incoming
  foreign-key constraints in \d, per discussion on hackers.  Also,
  take that string out of the translation roster, since it's now
  entirely pseudo-SQL.

Joe Conway committed:

- In pgsql/contrib/dblink/dblink.c, default client encoding to server
  encoding for dblink connections.  Addresses issue raised by
  Ruzsinszky Attila and confirmed by others.

Alvaro Herrera committed:

- In pgsql/src/backend/postmaster/autovacuum.c, dynamically set a
  lower bound on autovacuum nap time so that we don't rebuild the
  database list too often.  Per bug report from Lukasz Jagiello and
  ensuing discussion on pgsql-performance.

- In pgsql/src/backend/postmaster/autovacuum.c, fix typo, per Tom

Bruce Momjian committed:

- In pgsql/src/tools/pgindent/README, document new location for
  typedef list.

- In pgsql/src/tools/pgindent/README, update pgindent instructions.

- In pgsql/src/tools/find_typedef, fix BSD find_typedef logic.

- In pgsql/src/tools/find_typedef, fix Linux typedef code.

- In pgsql/src/tools/find_typedef, add FreeBSD support for
  find_typedef, per request from Andrew Dunstan.

- 8.4 pgindent run, with new combined Linux/FreeBSD/MinGW typedef list
  provided by Andrew Dunstan.

- In pgsql/src/tools/pgindent/pgindent, document struct/union problem
  with pgindent.

- In pgsql/src/tools/find_typedef, mention BSD/OS find_typedef

- In pgsql/src/tools/find_typedef, tighten typedef check for Linux.

Marc Fournier committed:

- Time to tag rc1 ...

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Kedar Potdar sent in a patch targeted for 8.5 to allow automated
range- and hash-based partitioning.

KaiGai Kohei sent in another revision of his SE-PostgreSQL patches.

Jeremy Kerr sent in three versions of a patch to refactor the
sigpipe-masking code in libpq from macros to private structs.

Kevin Grittner sent in a patch to make TOAST happen with less

Robert Haas sent in another revision of his patch refactoring EXPLAIN
for more generic output.

Robert Haas sent in another revision of his "generic options for
explain" patch.

Robert Haas sent in a patch to allow EXPLAIN to output in XML and
JSON.  It depends on the two patches above.

Robert Haas sent in a patch to have do_tup_output() take Datums.  This
is in aid of the EXPLAIN outputs above.

David Fetter and Petr Jelinek sent in patches to allow RETURNING in


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