> > As far as this question, seems with no RESTRICT/CASCADE, it fails, with
> > RESTRICT it drops the trigger, and with CASCADE it drops the referencing
> > table. Is that accurate?
>
> Not at all. CASCADE would drop the foreign key constraint (including
> the triggers that implement it), but not the other table. In my mind
> the issue is whether RESTRICT mode should do the same, or report an
> error.
>
> I'm not eager to accept the idea that DROP-without-either-option should
> behave in some intermediate fashion. I want it to be the same as
> RESTRICT.
I think that an unqualified drop should restrict and fail to drop if there's
a foreign key. Any app that lets people do a drop is probably already
checking for error conditions. Hence, it's just another error condition.
Chris