Zoltan Boszormenyi <zb@cybertec.at> writes:
> as the wishlist for PostgreSQL 8.4 still has my
> IDENTITY/GENERATED patches, I thought I refresh it.
> Before actually doing it though, I wanted to ask for opinions
> on implementation and ideas.
IIRC the end conclusion on that patch was that it was awfully messy,
didn't really buy much in new functionality, and was trying to track a
portion of the spec that the SQL committee had apparently not gotten
right yet.
> It seems the last draft for SQL:2008 largely clarified details for
> IDENTITY columns that were either conflicting or unclear in SQL:2003.
Implementing stuff that's still changing in a *draft* standard seems
to me like a good way to get burnt.
I think this needs to stay on the back burner, at least until there's
a published final spec that doesn't seem as broken as 2003 was. I
was sufficiently disillusioned with the whole thing at the end of
the previous review cycle
http://archives.postgresql.org/pgsql-hackers/2007-05/msg00441.php
that I'm not really eager to pour more time down the same hole.
regards, tom lane