[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Exactness

This page is part of the web mail archives of SRFI 77 from before July 7th, 2015. The new archives for SRFI 77 contain all messages, not just those from before July 7th, 2015.

bear scripsit:

> Instead of saying "rely on inexact numbers being floats", can
> you enumerate the qualities of floats that you find important?
> Because I think the right thing is to discuss (and possibly
> standardize on) the qualities, rather than a particular method
> of achieving them. IEEE floats are just one way of achieving
> those goals.

Let's be clear about what IEEE 754/854 is and is not.  There are
actually two standards: one prescribes *user-visible behavior*, the other
prescribes *representation in bits*.  Neither prescribes implementation,
and indeed there are many separate implementations.

Granted that a Scheme standard need not concern itself with representation
in bits, there is no reason for it to incorporate the whole of the
prescribed IEEE behavior.  Incorporation by reference is fine.  People who
do serious floating-point software rely on that set of behaviors and
will expect to find it if they can be persuaded to use Scheme at all.

Still less is there reason to incorporate the whole of the behavioral
prescription but with every "MUST" changed to "MAY at user option"
(figuratively speaking; the standard doesn't actually use RFC 2119

John Cowan                                   jcowan@xxxxxxxxxxxxxxxxx
        "You need a change: try Canada"  "You need a change: try China"
                --fortune cookies opened by a couple that I know