RI project status

Good news!

FYI, with contributions from Stephan Szabo and Don Baccus,
the FOREIGN KEY project made impressive progress during the
last days.

pg_dump is now able to dump FK constraints.

pg_dump arranges to disable/reenable all triggers during data
only reload.

ALTER TABLE ... ADD CONSTRAINT ... FOREIGN KEY is fully
implemented, and all existing data in the altered table is
verified to satisfy the new constraint.

The table actually created can be self referenced in the
constraints.

In contrast to my proposal, MATCH FULL and MATCH
<unspecified> will both be fully supported in 7.0 already. So
only MATCH PARTIAL will be left for 7.1.

The open items left for 7.0 are now the file buffering for
the trigger queue, the parser problem with NOT DEFERRABLE
(where Thomas actually jumps in), building a regression suite
and documentation.

Many thanks to the two guys above. Without them, FOREIGN KEY
would not only have failed to be finished in time. There
would have been a big mistake maken for NO ACTION at all,
leaving a huge hole for possible violations and not
conforming to the standard.

There is more to do after 7.0 is out, like ensuring that a
unique constraint is defined on referenced PK columns,
changing RESTRICT actions to fire as soon as possible and
ensuring uniqueness of constraint names. But what has been
done so far is IMHO really a major leap forward.

7.0 will have better FOREIGN KEY support than I expected.

Jan

--

#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#========================================= wieck(at)debis(dot)com (Jan Wieck) #