Re: Edge case conversions, exceptions to IEEE FPA
One more point I'd like to make in favour of intval(Inf)=Empty:
It preserves the relation isIn(x,intval(x)) for all floats x
(in the model where the underlying range is Non-extended Reals).
Now I'm not claiming that all such interesting relations favour
Arnold Neumaier's interpretation over Siegfried Rump's.
We should indeed come up with a list of such relations in one
place; many have been given, but they are scattered around.
Michel.
Sent: 2008-11-14 18:30:25 UTC