On Friday 10 November 2000 11:39, Tom Lane wrote:
> Peter Eisentraut <peter_e@gmx.net> writes:
> > When we implement schemas, then all objects belonging to the
> > DEFINITION_SCHEMA will not be dumped, all other objects will be. At
> > least I imagine that this might be something to work with.
>
> That's a thought, although it still doesn't cope with the issue of
> "what if I've altered a standard system object?" ... which is what
> I think Mark was getting at yesterday. I
Correct. I don't know why anyone would want to change the definition of
(say) int48eq, but if we are going to allow them to do so, we should be
careful to allow them to backup and restore such a change.
The template0 solution is at least better than what we have. And since I
have no other more brilliant suggestions, I would vote for it.
--
Mark Hollomon