Linux Today: Linux News On Internet Time.

More on LinuxToday

PostgreSQL Weekly News - July 30, 2003

Aug 01, 2003, 23:30 (0 Talkback[s])


How to Help Your Business Become an AI Early Adopter

[ Thanks to Robert Treat for this link. ]

PostgreSQL 7.3.4 rolled out without too many issues; by now it should be available from your local ftp mirror (full list of mirrors available at
http://www.postgresql.org/mirrors-ftp.html) in both source and rpm formats. The upgrade is critical for 7.3.3 users, and recommended for all others. If you would like to view the full release notes, they are available at http://www.postgresql.org/docs/release/734.html.

Meanwhile prep work for 7.4 beta is continuing. As there are still issues being worked through, it hasn't been packaged yet, since there is no sense packaging a product with known problems. Incidentally this works well for me as it gives me a good cliff hanger for each weeks newsletter. :-) Seriously though the first beta package should be available within the next week; in the mean time lets review this past weeks development.

Another (and perhaps final) round of error message editing was committed, thanks to the efforts of Tom Lane and Joe Conway. Areas that were finished up include the contrib modules, the utils, storage, and optimizer sections of the back end, and the various procedural languages. It's been noted that users of the various pieces of PostgreSQL should keep an eye on the error messages they receive once in beta to help ensure that the messages you receive are appropriate for the errors you trigger.

SSL client negotiation now has four different methods it can support; unencrypted non-SSL only, negotiated connection that prefer non-SSL, negotiated connections that prefer SSL, and SSL only connections. These changes work in conjunction with pg_hba.conf specific settings that can limit connections to the appropriate negotiated connection style. This work is thanks to Jon Jensen. Another SSL fix that went in was SSL text now printing only when SSL mode is enabled.

A number of time oriented fixes went in this week; timestamp_date for the HAVE_INT64_TIMESTAMP case was fixed, date_trunc('quarter',...) and it's ilk were corrected,and date/time input now follows along the lines of DateStyle rather than attempting to determine dates heuristically. Timetravel functionality also received some work, it now operates properly with "CREATE UNIQUE INDEX ixxx on table xxx", timetravel triggers no longer change the oid of the active record, and a get_timetravel(tablename) function was created to check timetravel status, along with some general cleanup of compile warnings.

In ecpg, the mdy function were fixed to use the correct offset, we now use yyless in place of the string_unput function, and some explicit casts for date, interval, and timestamp were added. JDBC received a few patches as well, fixing several sql injection attacks, thanks to Oliver Jowett and Dmitry Tkach.

A problem with pg_dumps handling of dropped columns was fixed, pg_dump now uses the same layout for types as it does for tables, and grant option dumping and some related cross-version compatibility issues in pg_dump where addressed. Psql now prints out a list of parent tables along with the standard table information display, thanks to Nick Barr, and psql's VERBOSE variable has been renamed to VERBOSITY.

While an official win32 build wont be coming in 7.4, there is still some work being done to improve windows support. The static library libpq.a did not work with psql, which is now fixed. Also password echoing has been disabled and 8bit characters are now properly displayed and interpreted. These fixes were all thanks to Christoph Dalitz.

Rounding out the weeks code changes are a slew of miscellaneous fixes including a number of outstanding portability issues with the new IPv6 code were tackled including using a more robust definition of struct sockaddr_storage, fixing the configure test to see if ss_len exists, no longer assuming getnameinfo() will handle AF_UNIX sockets, and more. Unknown-literal-constant default values will now be coerced to the column type during CREATE TABLE or ALTER TABLE SET DEFAULT, rather than postponing it to the time the default values is inserted into an INSERT command. The ERRCODE_XXX macros were moved to their own header file. Memory usage in the start-up process was tightened a little bit. A duplicate array in help_config.c was removed. Some general cleanup of the recent GUC changes. We now use a process-specific temp file name (based on suggestion by Andreas Haumer). An error in _bt_search was fixed. We now wait for the postmaster to actually exit when shutting down the postmaster after 'make check'. The --with-threads option was renamed to --enable-thread-safeness. The linux startup scripts were tweaked by Slawomir Sudnik. The initcap function has been made compatible with Oracle 9i. Error stack traceback support for was added for SQL-language functions. Pg_settings now displays the source of its information (default/user/session). And finally, the regression test for foreign key was improved to work with the deferred trigger code.

Next week we will round out the final code changes from feature freeze, and have more information on the start of beta testing.

PostgreSQL Product News

Mammoth PostgreSQL 7.3.3 Released

FUDforum 2.5.1 (Stable) released

pyPgSQL 2.4 released

PostgreSQL In the News

Gambling on Linux pays off in Nevada

PostgreSQL 7.3.4 Now Available For Download

rjjm.net blog entry with links to postgresql/jdbc information

Upcoming Events

Tcl'2003 BoF with PgAccess: Ann Arbor, Michigan: July 31 - Aug 2
A BoF will be held Friday night for PostgreSQL based tcl/tk projects

Linux World Expo: San Francisco, California: Aug 6-7
A BoF with the SF PostgreSQL Users group will take place Wednesday

PostgreSQL training at the Big Nerd Ranch: Atlanta, Georgia: Aug 18-22
Bruce Momjian will provide a one week training class.

1-day PostgreSQL Workshop: Turkey / Istanbul'da: August 23
FrontSITE & TDM is offering 1-day course on PostgreSQL

== PostgreSQL Weekly News - July 30th 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:

Related Stories: