> > Ah, I had this backwards: I thought SECURITY DEFINER wasn't
> > setting something that'd allow the foreign keys to run as the
> > owner of the function.
>
> Nah; by the time the RI triggers run, you're out of the function
> entirely. So they have to fend for themselves.
>
> Here's the 7.3 version of the patch (it's a bit ugly because I had
> to back-port a couple of changes that are in CVS tip).
Ah, excellent! Thank you Tom. This patch works wonderfully for me!
It doesn't apply cleanly to 7.2, but here's the diff in case anyone's
interested (applies to postgresql7 and postgresql-devel, just drop it
into files/ and recompile):
http://people.FreeBSD.org/~seanc/patches/patch_postgresql-7.3.2::src::backend::utils::adt::ri_triggers.c
-sc
--
Sean Chittenden