Bruce Momjian wrote:
>I am not sure what to do with this patch. It is missing dump
>capability, there is no clause to disable all triggers on a table, and
>it uses a table owner check when a super user check is required (because
>of referential integrity).
>
>
From a user's view, a trigger implementing RI isn't a trigger but an
implementation detail he shouldn't need to care about. So for std
triggers, owner check should be sufficient, requiring superuser for RI
triggers only.
This impacts EN/DISABLE TRIGGER ALL too. To touch RI triggers as well,
an additional keyword is needed.
Regards,
Andreas