> -----Original Message-----
> From: Doug Chamberlin [mailto:dchamberlin@...]
> Sent: Domingo 10 de Diciembre de 2000 8:56
> To: IB-Architect@egroups.com
>
> This enhancement request came through the sourceforge bug tracking system
> today. Remembering that NULL is not a value but a state, I'm not so sure
> this proposal is the right answer.
>
> I think you could make a case for having NULLs always trail the non-NULL
> values in a result set, regardless of the requested sort order of the
> result set.

With all respect, you took the report from other standpoint, Doug. It's not
about confusing values with non-values (NULL). It's not about applying a
collate to NULLs, too. The standard demands consistency, nothing more
complex. This is why I tagged it as a request for enhancement and not as a
bug. A bug is like that one Frank logged from a case in the NGs, where the
sorting process causes a lot of fields that have values to become NULL (see
second part of Bug #122376, I think the two parts of that bug need to be
decoupled).

The point is only: if you decided that in ASC order, NULLs should be
returned after values, in DESC order, NULLs should be returned before
values. I'm not discussing if the sort algorithm is stable or not.

C.