[plt-dev] Inexact integers

Michael Sperber <sperber at deinprogramm.de> writes:

> Jay McCarthy <jay.mccarthy at gmail.com> writes:
>
>> I'm totally ignorant of the history of this proposal, but I find it
>> very compelling. Was there any valid flak? Any improvements, etc?
>
> I'm biased, so it's hard for me to assess validity.  The most valid flak
> was, of course, that it would break backwards compatibility in a big
> way.  Also, you could no longer write straightforward programs that do
> some long-winded computation and afterwards make a decision based on
> whether the result is inexact or not.  I've personally never seen such a
> program.

There was also the idea that an inexact number may denote an interval -
and that IEEE floating-point is an implementation of that idea.  I think
this doesn't make sense: Of course, it's possible to view an IEEE number
as denoting the interval between the IEEE number just before it and the
one after it.  However, the IEEE operations aren't defined in terms of
those intervals: they are defined (simplifying somewhat) as operations
on "exact" numbers followed by rounding.

-- 
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla