Re: Query plans for plpgsql triggers

"Eric B. Ridge" <ebr@tcdi.com> writes:
> I should have mentioned that while the UPDATE statement in the
> trigger function really is as simple as the above, "some_other_table"
> is actually a view with the requisite ON UPDATE DO INSTEAD rule:

This is the sort of detail that you really should not omit.

If you were using something newer than 7.4 then I'd ask for a complete
test case so I could look into improving the behavior --- but as it is,
I'd first suggest upgrading and seeing if the problem is already fixed.

            regards, tom lane