> From: bear <xxxxxx@sonic.net>
> On Wed, 20 Jul 2005, Paul Schlie wrote:
>
>> Candidly, I don't any reasonable way for exact and inexact
>> representations to have the same infinite limit value, unless their
>> value ranges are correspondingly defined as being equivalent, which
>> typically isn't the case.
>
> Given the trouble it causes, perhaps they should be?

I wouldn't argue, as +/- 10^300 is enough for anything I can imagine doing.