> > But that requires a) dealing with the problem of the parent table
which has no
> > constraints and ...
Imho we should provide a mechanism that forces the parent to be empty
and let the planner know.
e.g. a false constraint on parent ONLY.
Andreas
