== PostgreSQL Weekly News - September 03 2006 == - Mailing list pgsql-announce

== PostgreSQL Weekly News - September 03 2006 ==

Peter Eisentraut starte a lively discussion on the functionality of
prepare.  His proposal was to change its functionality from "parse and
plan" to just "parse."  Other plans included making several plans,
caching them, and choosing an apprpriate one at run time.

Josh Berkus has kicked off the process of creating the 8.2 release
announcement.  It has resulted in some lively discussion and active
patching by Greg Sabino Mullane.

The Slony-I CVS archive has been unavailable all week due to the
outage of gBorg.  Downloads of source tarballs continue to be
available on pgFoundry, at
http://pgfoundry.org/frs/?group_id=1000122

The members of the Slony-I project anticipate a fourth release
candidate for version 1.2 soon after gBorg comes back online, and
apologize for any inconvenience.

== PostgreSQL Product News ==

Archiveopteryx 1.10 is out, with security fixes and new features.
http://www.archiveopteryx.org/1.10.html

Devrim Gunduz has been sponsored for Fedora Core packaging, and has
submitted python-psycopg2 to FC6, with more on the way.

PGCluster-1.0.11 released
http://pgfoundry.org/projects/pgcluster/

pgFouine is now in Debian.
http://packages.debian.org/unstable/misc/pgfouine

== PostgreSQL Local ==

PostgreSQL Turkey is now a legal organization, with Devrim Gunduz as
its first president.  Kudos!

Top French ISP Free now offers PostgreSQL for hosted sites.  Read on
for details in French.
http://www.freenews.fr/index.php?itemid=3742

== Applied Patches ==

Peter Eisentraut committed:

- A patch which repairs interpretation of GB as MB.

Tom Lane committed:

- A patch which fixes pgstat_report_waiting() to not dump core if
  called before pgstat_bestart() has been called.  It also makes
  pgstat_report_activity() safe against the same scenario, just in
  case.

- A partial fix for ecpg's VPATH problems.

- A speed optimization of trivial_subqueryscan() which makes it to
  consider a SubqueryScan's targetlist trivial if it contains either
  Vars referencing the corresponding subplan columns, or Consts
  equaling the corresponding subplan columns.  This lets the planner
  eliminate the SubqueryScan in some cases generated by
  generate_setop_tlist().

- A patch which adds an assign-hook mechanism for psql variables
  similar to the one existing for backend GUC variables, and use this
  to eliminate repeated fetching/parsing of psql variables in psql's
  inner loops.  Per discussion about how to ensure the upcoming
  FETCH_COUNT patch doesn't cost any performance when not being used.

- A patch from Chris Mair with revisions by Tom Lane which adds a
  FETCH_COUNT that causes psql to execute SELECT-like queries via a
  cursor, fetching a limited number of rows at a time and therefore
  not risking exhausting memory.

- A patch which shuts down the pager before reporting the error if a
  query fails part way through.

- A patch by Karel Zak and Zoltan Boszormenyi which extends COPY to
  support COPY (SELECT) TO [DESTINATION].  Kudos!

- A patch from Magnus Hagander which fixes things so that fopen's, not
  only open's, pass FILE_SHARE_DELETE and other special flags on
  Windows.  It may fix intermittent 'Permission denied' errors, too.

- A solution to the problem of making INSERT/UPDATE/DELETE ...
  RETURNING play nice with VIEWs and RULEs.

- A patch which cleans up a rather sloppy fix in HEAD for the ancient
  bug that CREATE CONVERSION didn't create a dependency from the new
  conversion to its schema.  He back-patched this to all supported
  releases.

- A Itagaki Takahiro/Tom Lane patch which arranges for GetSnapshotData
  to copy live-subtransaction XIDs from the PGPROC array into
  snapshots, and use this information to avoid visits to pg_subtrans
  in HeapTupleSatisfiesSnapshot.  This appears to solve the
  pg_subtrans-related context swap storm problem that's been reported
  by several people for 8.1.  While at it, modify GetSnapshotData to
  not take an exclusive lock on ProcArrayLock, as closer analysis
  shows that shared lock is always sufficient.

- A fix to description of archive_timeout.  archive_command is only
  invoked on completed WAL segments.

Teodor Sigaev committed:

- A patch which fixes a bug where building a GIN index on an empty
  table could crash the server.

- A patch which adds a description of tsvector's interal structure to
  its header file (tsvector.h).

- A patch which removes a non-useful code path in silly_cmp_tsvector()

- A patch which fixes an incorrect calculation of the length of
  lexemes in silly_cmp_tsvector()

Michael Meskes committed:

- Another patch intended to fix ecpg's regression tests on
  VPATH-enabled machines.

- A patch which removes the debug option from pg_regress in ecpg.

- A patch which fixes ecpg's parser and library to allow empty
  database names, streamlines connection name parsing and adds
  Joachim Wieland's patch to shorten paths before diffing.

Bruce Momjian committed:

- A patch per Guillaume Smet's patch which makes log_statement
  prepares easier to read.

- A patch per Michael Glaeseman which fixes some edge cases in date
  arithmetic.

- Susan Ebrecht's patch which adds a new SQL standard idiom for
  updating multiple columns at once:
    UPDATE table SET (col, ...) = (val, ...)

- A patch from Greg Sabino Mullane with changes per Tom Lane which
  adds a new server_version_num variable which presents the server
  version as in integer, which in turn makes it easy to do comparisons.

- Satoshi Nagayasu's patch which adds functions to contrib/pgstattuple
  that show index statistics and index page contents.

- Sven Suursoho's patch which allows PL/Python to return composite
  types and result sets (i.e. SETOF)

- A patch from Peter Eisentraut that makes autovacuum behavior more
  aggressive, per discussion on hackers list.  This was originally
  part of "autovacuum defaults to 'on'" patch that was reverted.

- A patch which removes an unnecessary copyObject() call in Susan
  Ebrecht's UPDATE (values) code.

- A patch which fixes the case where "PM" to_timestamp() mask was
  using up too many characters.  Per report from Josh Tolley.

Joe Conway committed:

- A patch originally by Kai Londenberg for dblink which adds async
  query capability.

== Rejected Patches (for now) ==

- Euler Taveira de Oliveira's patch which suppresses some NOTICE
  messages from REINDEX command.  Need to look into needs/consistency.

- Dhanaraj M's patch which changes FETCH and MOVE to use int8.

== Pending Patches ==

ITAGAKI Takahiro sent in a new version of his patch meant to address
the context-switch storm problem which affects multi-processor
PostgreSQL installations.

Bernd Helmle sent in another version of his updatable view patch whose
functions use expression_tree_walker()

Christopher B. Browne sent in a patch which fixes the ecpg regression
tests on AIX.

Zdenek Kotala sent in a documentation patch which extends the
documentation for the system catalog tables.

Victor B. Wagner sent in a patch, then a revision for same, which
improves OpenSSL support both for current and upcoming versions of
OpenSSL including two new GUC variables: ssl_ciphers, which allows the
server administrator to  specify set of SSL ciphersuites which can be
used by clients to connect to the server, and ssl_engine, which allows
you to specify a loadable crypto engine (i.e.  hardware crypto
accelerator support) to use.

Tom Dunstan sent in the first version of his enums patch.

== PostgreSQL in the News ==

Planet PostgreSQL: http://www.planetpostgresql.org/

General Bits, Archives and occasional new articles:
http://www.varlena.com/GeneralBits/

PostgreSQL Weekly News is brought to you this week by David Fetter
and Dave Page.