"Rolf A. de By" <deby(at)itc(dot)nl> writes:
> Thanks for that. There is some misunderstanding here. For this example,
> I had taken the sting out of my trigger function and turned it into a
> much more concise no-op, with warnings. The actual code of my original
> trigger function is irrelevant. The no-op trigger function displays the
> same strange behaviour: it works as expected for INSERTs, but not for
> UPDATEs. The update goes through! And it shouldn't.
Reading between the lines, I gather you have an inheritance setup and
are expecting a trigger on the parent table to fire for events occurring
in the child tables. Doesn't work like that; you need to put triggers
on the child tables.
regards, tom lane