"Jonathan Ellis" <jonathan@utahpython.org> writes:
> I can do that... you don't think the fact I mentioned, that
> redefining the view to leave out the expensive function fixes the
> problem, is relevant?
Hm, I'd not have thought that an expensive function would get evaluated
partway up the join tree, but maybe that's wrong. You never did show
us the actual view definition ...
regards, tom lane