Re: Deferrable UNIQUE INDEX?

On Tue, Jan 03, 2006 at 11:17:16PM -0500, Tom Lane wrote:
> "Andrew Dunstan" <andrew@dunslane.net> writes:
> > David Fetter said:
> >> Is there some way to make the index check INITIALLY DEFERRABLE
> >> the way a regular column/table constraint could be?
> 
> > The TODO list has: "Allow DEFERRABLE UNIQUE constraints?" send in
> > a patch ;-)
> 
> This is definitely on the wish-list, because it is both useful and
> required by the SQL spec ... but it seems nontrivial.  I recall some
> discussions in the pghackers archives about how to do it.

There was some discussion in the followups to this posting:

http://archives.postgresql.org/pgsql-hackers/2005-01/msg00882.php

Alvaro mentioned a strategy here

http://archives.postgresql.org/pgsql-hackers/2005-01/msg00904.php

and the "tough part" appears to have come down to "disk-spilling
logic."

I'm out of my depth here, but maybe we could figure out just what
components would need to be touched and how...

Cheers,
D
-- 
David Fetter david@fetter.org http://fetter.org/
phone: +1 415 235 3778

Remember to vote!