== PostgreSQL Weekly News - August 13th 2003 == - Mailing list pgsql-announce

== PostgreSQL Weekly News - August 13th 2003 ==

    The beta process marches on this week with the hot topics centering
around various compiling issues on different hardware and operating
systems. It is interesting to watch some of the folks we haven't seen in
a bit now popping up with things to be fixed in the beta process. While
things have been pretty smooth so far, we'll see what shakes out in the
next week or two.

    Miscellaneous fixes this week were pretty wide ranging, but we'll start
with the ones that are build related. A compiler-detected problem for
Alphas was fixed where strlen would return something wider than int.
Along with this a faulty assumption that sizeof("INT_MAX") had something
to do with the maximum number of digits in an int was corrected. The
fields of DestReciever were renamed to avoid collisions with macros in
some platforms sys/socket.h. Some unnecessary uses of multiple cat
processes to feed psql were removed, reducing the  number of concurrent
processes launched during parallel regression tests and possibly
avoiding failures such as exceeding a user's max number of processes.
After some problems with AIX and Unixware, we no longer  use
HAVE_STRUCT_ADDRINFO as a guide to whether netdb.h defines macros like
AI_NUMERICHOST, instead the macros are tested individually. We also no
longer assume that the struct option is available just because we can
find a getopt_long() since some user-added libraries contain the
subroutine, but there is no guarantee we will find the matching header
files. Finally we no longer link in libwsock32 on non-win32 platforms.

    The create_unique_plan function no longer discards existing output
columns of the subplan it starts with, as they may be needed at upper
join levels. After some discussion on the -hackers mailing list, a
random back off algorithm for sleep delays was adopted when waiting for
a spinlock. Floating-point timestamp comparisons were fixed n cases
where NaN is encountered based on a bug report from Christian van der
Leeden. The larger/smaller routines (MAX/MIN) were also adjusted to
share code with comparisons for timestamp, interval, and timetz. The
space between function name and the opening paren was removed for added
clarity in pg_dump, and psql now uses the new pg_get_viewdef function
for improved formatting. The file threads.c was renamed to thread.c,
simple_prompt()/spromt.c was moved into the /port directory, and null
checking of type decimal datums without indicator in Informix mode were
fixed. The conversion function were all set to strict to prevent them
from getting NULL inputs. Unused variable warnings are now suppressed
when building without Asserts. The use of malloc(0) is now avoided in
plpgsql as some debugging malloc packages object to it.

    A bug report from Mendola Gaetano show an order-of-operations bug in
_SPI_cursor_operation which was fixed. A potential deadlock situation
created by recent changes to recycle btree index pages was repaired. A
general code clean up was done in response to a recent resname bug
report, including making sure the rewriter and planner only use resno
and not resname to identify targets in INSERT and UPDATE targetlists.
The search loop in pgstat_initstats was tightened and we now avoid some
unnecessary work when stats collection is disabled. Some
frequently-reached elog(DEBUG...) calls were changed to
ereport(DEBUG...) for speed reasons. We can now cope with NoData message
from backed (needed for case where PQexecParams is used with a statement
that doesn't return data). And of course we had more doc updates,
including enhancements to 7.4 array features and polymorphic functions,
better reference links within the manual, and some notes on using
OpenFTS vs. Tsearch2.

    The JDBC folks made a number of changes this week, including several
patches from Oliver Jowett to improve buffer sizing, cleanup exception
handling,  fix a deprecation in newer versions of ant, better handle
invalid input for getArray, and cleanup documentation and the README
file. The regression tests were tweaked to work with different responses
to transaction isolation level between version of PostgreSQL.  A problem
with trying to use a fetch when a cursor wasn't being used was
fixed, that patch being sent in from kho@redhat.com. Another go around
of sql injection fixes took place, this time the focus was on preventing
modification of the where clause on the current statement or performing
a subselect which could circumvent application security logic. A few
other documentation updates were added, including information on calling
stored functions.

    ECPG also was back in swing this week. The makefile was modified so -lm
is no longer required for ecpg compiles. The ecpg thread test program
was tweaked to be more automated, and an iteration option was added.
There were also some doc cleanups done as well.

    That's it for this weeks development, though I do have an
infrastructure note; the mail2news gateway should be fully functional,
so those who have been struggling trying to follow things through the
newsgroups should be back in business now.



== PostgreSQL Product News ==

Postgresql AutoDoc 1.12 Released
http://freshmeat.net/releases/132367/

Bricolage 1.6.4 Released
http://archives.postgresql.org/pgsql-announce/2003-08/msg00015.php



== PostgreSQL In the News ==

Govt agency learns a lesson in open source
http://www.computerworld.com.au/index.php?id=298405703&fp=16&fpid=0

'Safe' Perspective on Interoperability and Open Source GIS
http://www.directionsmag.com/article.php?article_id=406

Using PostgreSQL with Cold Fusion (blog)
http://www.redev.org/archives/000023.cfm#more



== Upcoming Events ==

PostgreSQL training at the Big Nerd Ranch : Atlanta, Georgia: Aug 18-22
Bruce Momjian will provide a one week training class.
http://www.bignerdranch.com/Classes/Postgresql.html

1-day PostgreSQL Workshop : Turkey / Istanbul'da : Aug 23
FrontSITE & TDM is offering 1-day course on PostgreSQL
http://www.frontsite.com.tr and
http://www.tdmsoft.com/egitim_ayrinti.php?id=1

Gnu/Linux Congress: Veracruz, Mexico: September 17-19
Bruce Momjian has been asked to give a keynote address and he will also
be presenting tutorials
http://congreso.gulev.org.mx



== PostgreSQL Weekly News - August 13th 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