At 05:33 PM 28/10/2004 +1000, you wrote:

> >
> > The nullable character of the domain, the "one-null" rule of
> > the unique
> > constraint and the rules for foreign key indexes allow this
> > orphan row to
> > happen. It think it needs to be reported as a bug (a
> > possible side-effect
> > from the changes to the uniqueness rules in 1.5). Until or
> > unless it is
> > corrected, beware of it - don't allow nulls on foreign key
> > columns unless
> > you've a reason to want them there, i.e. you plan to implement the ON
> > DELETE | UPDATE SET NULL action rule.
>
>
>Will you report this as a bug, or should I, as I am not sure I can consider
>this a bug.

It's not a bug for you, then, is it? It's not a bug for me, either, since
I'd never want to use RI to implement a rule like this. It's logically
wrong, which means someone is probably gonna want to fix it some day. In
the meantime, it remains a handy trick for you and a must-to-avoid for me,
but not a bug until someone gets burnt by it and wants it fixed.

./hb