Обсуждение: Re: [HACKERS] Group By, NULL values and inconsistent behaviour.

> > >For now, ignore the patch I sent.  Appears from Andreas demo that the
> > >current postgres code will follow the Informix style with regard to
> > >grouping columns with NULL values. Now that I really think about it,
> > >it does make more sense.
> >
> > I think I saw the patch committed this morning...?
>
>     Yesterday evening, actually...should we back it out, or leave it
> as is?  Is the old way more corrrect the the new?

Marc,

From the two responses demonstrating Informix and Sybase, I think the patch
I sent should be backed out.  I can live with NULL equaling NULL in the
GROUP BY and ORDER BY clauses, but not in the WHERE clause, if everyone else
is doing it that way.

darrenk

>
> > > >For now, ignore the patch I sent.  Appears from Andreas demo that the
> > > >current postgres code will follow the Informix style with regard to
> > > >grouping columns with NULL values. Now that I really think about it,
> > > >it does make more sense.
> > >
> > > I think I saw the patch committed this morning...?
> >
> >     Yesterday evening, actually...should we back it out, or leave it
> > as is?  Is the old way more corrrect the the new?
>
> Marc,
>
> >From the two responses demonstrating Informix and Sybase, I think the patch
> I sent should be backed out.  I can live with NULL equaling NULL in the
> GROUP BY and ORDER BY clauses, but not in the WHERE clause, if everyone else
> is doing it that way.
>
> darrenk
>
>

Your patch backed out.

--
Bruce Momjian
maillist@candle.pha.pa.us