On Wed, 19 Dec 2001, Tom Lane wrote:
> "Ian Harding" <ianh@tpchd.org> writes:
> > Is this something that will change anytime soon?
>
> It's not high-priority in my mind. The difficulty is the lack of a
> planning step for constraint expressions. Maybe we could fix this
> when we redesign querytrees; right now I think there's too much cruft
> in the way (memory leaks, state kept in the querytree, etc etc)
Also, AFAICS, check constraints with subselects are constraints on the
other tables named in the subselect as well and it could be on a
case which the current check constraints don't fire (for example
on delete from a table where a set value function is used, like
check (select count(*) from foo)>col1) or something like that).