Skip to content

Wöchentlicher PostgreSQL Newsletter - 27. Februar 2011

Der Originalartikel befindet sich unter:

== Wöchentlicher PostgreSQL Newsletter - 27. Februar 2011 ==

Gabriele Bartolini von der ITPUG wird einen Vortrag über
PostgreSQL 9.0 auf der Codemotion in Rom am Samstag dem 5. März 2011
um 16:10 Uhr halten. Mehr Informationen unter:

== PostgreSQL Produkt Neuigkeiten ==

pgpool-II 3.0.3, ein Connection Pooler und mehr,
ist erschienen.

pgwatch 1.0beta1, ein Monitoringwerkzeug für PostgreSQL,
ist erschienen.

psycopg2 2.4, ein Python Treiber für PostgreSQL, ist erschienen.

== PostgreSQL Jobs im Februar ==

== PostgreSQL Lokal ==

PostgreSQL Conference East 2011: New York,
vom 22. bis 25. März

Das Open Database Camp findet vom 7. bis 9. Mai 2011 in Sardinien,
Italien statt.

PGCon findet am 19. und 20. Mai 2011 an der Universität
von Ottawa statt, vorher gibt es am 17. und 18. Mai
zwei Tage mit Trainings.

== PostgreSQL in den News ==

Planet PostgreSQL:

Dieser wöchentliche PostgreSQL Newsletter wurde erstellt von David Fetter.

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

== Reviews ==

== Angewandte Patches ==

ITAGAKI Takahiro pushed:

- Add ENCODING option to COPY TO/FROM and file_fdw.  File encodings
  can be specified separately from client encoding.  If not specified,
  client encoding is used for backward compatibility.  Cases when the
  encoding doesn't match client encoding are slower than matched cases
  because we don't have conversion procs for other encodings.
  Performance improvement would be be a future work.  Original patch
  by Hitoshi Harada, and modified by me.

- Fix pg_server_to_client, that was broken in the previous commit.

- Make the second words lowercase in psql's \d titles for unlogged

- Add tab-completion for CREATE UNLOGGED TABLE in psql, and fix
  unexpected completion for DROP TEMP and UNIQUE.

- More psql tab-completion for new commands:  ALTER FOREIGN DATA

Bruce Momjian pushed:

- Update docs about new role privileges reference, per Alvaro Herrera.

- Move information_schema duplicate constraint note to the top of the
  information schema documentation because it affects several tables.

- Reword information_schema duplicate constraint warning text.

- Update wording about information schema and name which views
  potentially can have duplicates, per request from Tom Lane.

- Document pg_options_to_table() (not previously documented)

- Fix doc patch --- pg_options_to_table() returns "setof record".

- Document that pg_options_to_table() also works for

- Fix markup for pg_options_to_table() to report the return column
  names, per suggestion from Andrew Dunstan.

- Be less detailed about reporting shared memory failure by avoiding
  the output of actual Postgres parameter values related to shared
  memory, and suggesting that these are only possible parameters to

- Document that last vacuum statistics and counts are for non-FULL

Tom Lane pushed:

- Fix dangling-pointer problem in before-row update trigger
  processing.  ExecUpdate checked for whether ExecBRUpdateTriggers had
  returned a new tuple value by seeing if the returned tuple was
  pointer-equal to the old one.  But the "old one" was in
  estate->es_junkFilter's result slot, which would be scribbled on if
  we had done an EvalPlanQual update in response to a concurrent
  update of the target tuple; therefore we were comparing a dangling
  pointer to a live one.  Given the right set of circumstances we
  could get a false match, resulting in not forcing the tuple to be
  stored in the slot we thought it was stored in.  In the case
  reported by Maxim Boguk in bug #5798, this led to "cannot extract
  system attribute from virtual tuple" failures when trying to do
  "RETURNING ctid".  I believe there is a very-low-probability chance
  of more serious errors, such as generating incorrect index entries
  based on the original rather than the trigger-modified version of
  the row.  In HEAD, change all of ExecBRInsertTriggers,
  ExecIRInsertTriggers, ExecBRUpdateTriggers, and ExecIRUpdateTriggers
  so that they continue to have similar APIs.  In the back branches I
  just changed ExecBRUpdateTriggers, since there is no bug in the
  ExecBRInsertTriggers case.

- Remove ExecRemoveJunk(), which is no longer used anywhere.  This was
  a leftover from the pre-8.1 design of junkfilters.  It doesn't seem
  to have any reason to live, since it's merely a combination of two
  easy function calls, and not a well-designed combination at that (it
  encourages callers to leak the result tuple).

- Allow binary I/O of type "void".  void_send is useful for the same
  reason that void_out doesn't throw error, namely that someone might
  do "select void_returning_func(...)"  from a client that prefers to
  operate in binary mode.  The void_recv function may or may not have
  any practical use, but we provide it for symmetry.  Radosław Smogura

- Add a relkind field to RangeTblEntry to avoid some syscache lookups.
  The recent additions for FDW support required checking
  foreign-table-ness in several places in the parse/plan chain.  While
  it's not clear whether that would really result in a noticeable
  slowdown, it seems best to avoid any performance risk by keeping a
  copy of the relation's relkind in RangeTblEntry.  That might have
  some other uses later, anyway.  Per discussion.

- Support data-modifying commands (INSERT/UPDATE/DELETE) in WITH.
  This patch implements data-modifying WITH queries according to the
  semantics that the updates all happen with the same command counter
  value, and in an unspecified order.  Therefore one WITH clause can't
  see the effects of another, nor can the outer query see the effects
  other than through the RETURNING values.  And attempts to do
  conflicting updates will have unpredictable results.  We'll need to
  document all that.  This commit just fixes the code; documentation
  updates are waiting on author.  Marko Tiikkaja and Hitoshi Harada

- Fix order of shutdown processing when CTEs contain inter-references.
  We need ExecutorEnd to run the ModifyTable nodes to completion in
  reverse order of initialization, not forward order.  Easily done by
  constructing the list back-to-front.

- Refactor the executor's API to support data-modifying CTEs better.
  The originally committed patch for modifying CTEs didn't interact
  well with EXPLAIN, as noted by myself, and also had corner-case
  problems with triggers, as noted by Dean Rasheed.  Those problems
  show it is really not practical for ExecutorEnd to call any
  user-defined code; so split the cleanup duties out into a new
  function ExecutorFinish, which must be called between the last
  ExecutorRun call and ExecutorEnd.  Some Asserts have been added to
  these functions to help verify correct usage.  It is no longer
  necessary for callers of the executor to call
  AfterTriggerBeginQuery/AfterTriggerEndQuery for themselves, as this
  is now done by ExecutorStart/ExecutorFinish respectively.  If you
  really need to suppress that and do it for yourself, pass
  EXEC_FLAG_SKIP_TRIGGERS to ExecutorStart.  Also, refactor portal
  commit processing to allow for the possibility that PortalDrop will
  invoke user-defined code.  I think this is not actually necessary
  just yet, since the portal-execution-strategy logic forces any
  non-pure-SELECT query to be run to completion before we will
  consider committing.  But it seems like good future-proofing.

- Add documentation for data-modifying statements in WITH clauses.
  Marko Tiikkaja, somewhat reworked by Tom Lane.

Robert Haas pushed:

- Fix a couple of unlogged tables goofs.  "SELECT ... INTO UNLOGGED
  tabname" works, but wasn't documented; CREATE UNLOGGED SEQUENCE and
  CREATE UNLOGGED VIEW failed an assertion, instead of throwing a
  sensible error.  Latter issue reported by Itagaki Takahiro; patch
  review by Tom Lane.

- Named restore point improvements.  Emit a log message when creating
  a named restore point, and improve documentation for
  pg_create_restore_point().  Euler Taveira de Oliveira,  per
  suggestions from Thom Brown, with some additional wordsmithing by

Peter Eisentraut pushed:

- Add PL/Python functions for quoting strings.  Add functions
  plpy.quote_ident, plpy.quote_literal, plpy.quote_nullable, which
  wrap the equivalent SQL functions.  To be able to propagate char *
  constness properly, make the argument of quote_literal_cstr() const
  char *.  This also makes it more consistent with quote_identifier().
  Jan Urbański, reviewed by Hitoshi Harada, some refinements by Peter

- Table function support for PL/Python.  This allows functions with
  multiple OUT parameters returning both one or multiple records
  (RECORD or SETOF RECORD).  Jan Urbański, reviewed by Hitoshi Harada

- Remove remaining expected file for Python 2.2.  We don't have
  complete expected coverage for Python 2.2 anyway, so it doesn't seem
  worth keeping this one around that no one appears to be updating
  anyway.  Visual inspection of the differences ought to be good
  enough for those few who care about this obsolete Python version.

- PL/Python explicit subtransactions.  Adds a context manager,
  obtainable by plpy.subtransaction(), to run a group of statements in
  a subtransaction.  Jan Urbański, reviewed by Steve Singer,
  additional scribbling by me

Alvaro Herrera pushed:

- Fix pageinspect's heap_page_item to return infomasks as 32 bit
  values.  HeapTupleHeader's t_infomask and t_infomask2 are defined as
  16-bit unsigned integers, so when the 16th bit was set,
  heap_page_item was returning them as negative values, which was
  ugly.  The change to pageinspect--unpackaged--1.0.sql allows a
  module upgraded from 9.0 to be cleanly updated from the previous

- Update pageinspect--1.0.sql to match the upgrade script.  Per
  comment from Tom Lane

Andrew Dunstan pushed:

- Add a configure check for Python.h if building with python.

Heikki Linnakangas pushed:

- Increase the default for wal_sender_delay from 200ms to 1s.  Now
  that WAL sender is immediately woken up by transaction commit,
  there's no need to wake up so aggressively.

Magnus Hagander pushed:

- Fix verbose display of REPLICATION role attribute.  Josh Kupershmidt

== Abgelehnte Patches (bis jetzt) ==

No one was disappointed this week :-)

== Eingesandte Patches ==

Hans-Juergen Schoenig sent in a WIP patch to track cross-column

Kevin Grittner sent in two revisions of a patch to track changes with

Radoslaw Smogura sent in three revisions of a patch to add a binary
in/out for aclitem.

Alexander Korotkov sent in a WIP patch to collect statistics for array

Shigeru HANADA sent in another revision of the postgresql_fdw patch.

Gurjeet Singh sent in a patch to allow the use of paths relative to
that of the current file in psql.

Andrew Tipton sent in a patch to add GiST support for BOX @> POINT

Magnus Hagander sent in another revision of the patch to allow
streaming a base backup.

Fujii Masao sent in another revision of the patch to add replication
server timeout.

Rumko (rumcic AT gmail DOT com) sent in a patch to allow PostgreSQL to
compile on Dragonfly BSD.

Jaime Casanova sent in another revision of the patch to allow for
synchronous replication.


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