> > it, or you make up arbitrary and potentially inconsistant
results.
> > You could say, of course, that once you make the results
> consistant,
> > it's a rule (which you may or may not want to apply every time).
> >
I also meant to say I think the consistency is a feature rather than a bug; it emans you can model many minor things that are not in themsleves worth a rule.
Powered by hypermail