On Wed, 2009-01-21 at 17:39 -0500, Bruce Momjian wrote:
> > It looks like it would be easy enough to throw a better error message
> > than that, e.g. with a try/catch. The information could be obsolete, but
> > if it succeeds, it would at least mean they had permissions at some time
> > in the past.
> >
> > Or, we could just remove the ACL checks from LOCK TABLE, so that it's at
> > least consistent. Mostly it's the inconsistency that bothers me.
>
> Is this a TODO?
I don't feel too strongly about it. I would feel better if we were
consistent about the permissions checks, because there's less of a
chance for confusion or a false sense of security.
If we keep the permission check in LockTableCommand(), I can make a
patch that produces a more useful error message when the table is
removed right before the pg_class_aclcheck().
Right now it does:
ERROR: relation with OID 16542 does not exist
which is undesirable.
Regards,Jeff Davis