== PostgreSQL Weekly News - September 18th 2003 ==
On Monday 7.4 beta3 was packaged up, if you haven't given it a spin it
should be available from your local mirror. We still saw a number of
issues crop up so every extra person who tries a beta package brings us
that much closer to a final release.
A lot of debate centered around how best to ensure use of our spinlock
code on all platforms after it was noticed that PostgreSQL would compile
without using it on some architectures, which leads to a significant
slow down of the database. One way to test if your database built
without spinlocks would be to check "ipcs -s" and count up the number of
semaphores created by the postmaster. Normally PostgreSQL will grab
approximately max_connections semaphores, but without spinlocks it would
be somewhere above max_connections + 2 * shared_buffers. For 7.4 a
compile error will be thrown if spinlock code is not found, and a
configure flag (--disable-spinlocks) will be needed to get past this
error. Should you run across this problem installing one of the beta
packages, please report it along with your OS and Hardware information
to the pgsql-hackers mailing list.
General fixes this week included squaring away some uppercase type
definitions in ecpg, more tweaking on reentrent functions and thread
support, adding a hint to the CREATE FUNCTION command for known
languages that may not be loaded, adjusting copydir() to emit a WARNING
rather than ERROR on failure, psql should now invoke the pager when
calling for \? and sql help in psql is now translatable, a missed
message was updated for opclasses, and a bug in timetravel was fixed.
Several warnings were removed from commands that have no effect when
executed repeatedly. A race condition reported by Gavin Sherry for
LISTEN/NOTIFY was patched until a more general fix can be derived.
SPI_modifytuple now accepts a zero for the natts argument (implying no
changes to make). The use of mktime on the back end was eliminated in
favor of localtime due to a number of bugs that kept coming up. A fix
for aset.c unnecessarily making malloc() requests of sizes that aren't
power of 2. The MemoryContextCheck() call was moved to after commit to
avoid duplication of effort. Plpython now generates separate cached
procedure data for each relation when the same function is used as a
trigger on more than one relation and some issues with plpython's
dropped column were worked out.
Several contrib modules saw work this week as well. The srandom() call
was removed from normal_rand() in the tablefunc module. The cube module
was updated to work with flex 2.5.31, and now uses a real btree operator
class. The dblink regression tests were updated. An incorrect pfree in
to_ts_query_name() of the tsearch2 package was fixed, and it also was
updated to work with flex 2.5.31. The new pg_autovacuum feature saw more
updates this week. Quotes were added to vaccum/analyze commands to
handle upper/lowercase object names. There were also some changes put in
to help work through some issues that came up with the stats system (for
which a new regression test was added). On another note the
contrib/array was mostly removed with explanation on the new syntax for
achieving these types of operations.
Thanks to the folks at Red Hat there were several more fixes put in for
the JDBC driver this week. Patches contained more SQLState updates
including correctly reporting SQLState support, improvements to boolean
and bit handling, and some corrections for error messages that were not
translating correctly.
As usual we round out the weeks changes with what seemed like an
especially high number of documentation changes. Areas that saw changes
included the create function page, a datetime appendix table, general
function information, the run time config sections, the client
authentication section (along with corresponding notes in the
pg_hba.conf.sample), information schema, and the analyze page. GUC
variable names should now be consistently lowercase throughout the docs,
several outdated error messages were cleaned up, several sgml
conventions were made more consistent, and there was more cleanup
of the diagnostic sections.
== PostgreSQL Product News ==
Lyris ListManager 7.5 Available
http://www.marketwire.com/mw/release_html_b1?release_id=57715
Safe Software Adds PostGIS & PostgreSQL Reading & Writing Support to FME
http://www.directionsmag.com/press.releases/?duty=Show&id=7852
EMS PostgreSQL Manager 1.8 released
http://www.databasejournal.com/news/article.php/3075321
== PostgreSQL In the News ==
RedSheriff transacts to a lower TCO
http://www.computerworld.com.au/index.php?id=1482975508&fp=16&fpid=0
Recursive constraint solved! (blog)
http://frontierj.blogspot.com/2003_09_14_frontierj_archive.html#106365830849624470
== Upcoming Events ==
LinuxWorld Conference & Expo 2003 : Frankfurt, Germany: October 27-29
PostgreSQL will have a booth at the LinuxWorld Conference
http://www.linuxworldexpo.de/
International PHP Conference: Frankfurt, Germany: November 2-5
Bruce Momjian will be presenting several PostgreSQL talks at this year's
International PHP Conference.
http://www.phpconference.de/2003/
NordU Usenix 2004: Denmark: January 28 - February 1
Bruce Momjian will be giving PostgreSQL several tutorials
http://www.nordu.org/NordU2004/
== PostgreSQL Weekly News - September 18th 2003 ==
Don't forget to read Elein Mustain's Weekly Summary of the PostgreSQL
General Mailing List http://www.varlena.com/GeneralBits/
On the Web:
http://www.postgresql.org
http://advocacy.postgresql.org
