Dear Tom,
> > The standard "EVERY" is fine for postgres, the issue is only with
> > ANY/SOME. Do you think that bool_and should be proposed anyway for
> > homogeneity with bool_or?
>
> I think EVERY is actively misleading, because it does *not* imply that
> every input is TRUE. The spec says these aggregates should ignore
> nulls, and so a true result only implies that there were no FALSE
> inputs.
>
> OTOH one could argue that the ignore-nulls behavior makes this not a
> true analog of AND, either ...
Argh, how stupid I am, I missread the specification! Then the patch I sent
yesterday is wrong if NULL values are encountered:-( I should learn how to
read sometimes...
However, I did not name the boolean and aggregate EVERY, it is BOOL_AND
(in the patch), because I tend to prefer homogeneity.
I'll resubmit a patch later.
--
Fabien Coelho - coelho@cri.ensmp.fr
