"Tommy McDaniel" <tommstein(at)myway(dot)com> writes:
> Let us also create a trigger to disable UPDATEs on table_2:
> ...
> And, we have now broken referential integrity.
Yup, this is not a bug, it's a feature. Triggers fire on
referential-integrity updates. (If they didn't, you could not for
example have a logging trigger log RI actions.) If you don't want
to break RI, you'd better think more carefully about what your
trigger does.
regards, tom lane