> The solution to the foreign key problem seems easy if I
> modify PostgreSQL implementation and take off the ONLY word
> from the SELECT query, but it's not an option for me, as I'm
I think that the ONLY was wrong from day one :-(
The default in other areas is table including childs.
(Not like in old pg where you had to use tab* to include childs)
(iirc leaving off ONLY is not sufficient because of locking
problems)
Of course then we would need
REFERENCES tenk ONLY (unique1)
to allow current behavior.
Andreas