On Wed, 31 Oct 2001, Aasmund Midttun Godal wrote:
> Well, the foreign keys are dumped out as separate statements, so I
> think it will work. It's definitely worth a try...
It'll only write out part of the constraint in those cases I believe
(and it seems to be borne out by my 7.2 pg_dump output). It looks like it
writes out the triggers for the table that's specified, but not for the
other side of the constraint, but when you do the drop table, all of the
constraints are removed. The restore after a drop table would probably
be a partially functioning constraint.