The current proposal
> NO, whether intervals should contain Infinity as a member
> is NOT the point!
Thanks for correcting my misconception, which I was led to by your
examples of direct arithmetic on interval bounds (which may include
Inf in perhaps all versions of IA).
So the sticky point is conversion of the scalar Inf to an interval,
which was discussed at length in the past, and which was addressed
again with several proposals earlier today.
> If there is no acceptable way to maintain all properties (1..4)
> we may think about workarounds and look for the best compromise.
> Of course, the speed is important as well.
As Arnold Neumaier and others have pointed out, we have to be careful
for what we wish for. But I agree that listing such requirements for
the purpose of examining both implications and feasibility is a very
useful excercise.
Michel.
Sent: 2009-02-23 16:10:00 UTC