Re: reading NaNs bear 27 Oct 2005 05:47 UTC


On Wed, 26 Oct 2005, Bradd W. Szonye wrote:

> I would hope that anyone reviewing IEEE-754 features for Scheme is
> already familiar with the revision and its current status. It'd be silly
> to adopt a 20-year-old standard just a few months before a major
> revision. Then again, it might be silly to wait on the new draft, given
> its troubles. Either way, arguments should at least take it into
> account.

I would not say that I'm "familiar" with it, but I have heard
noises about it.

http://grouper.ieee.org/groups/754/revision.html

is a good resource.

http://en.wikipedia.org/wiki/IEEE_754r

Is another one.

They intend to standardize new floating point formats including
3 sizes of decimal and new 128-bit binary floats, be more
specific about exceptions and handlers for those exceptions,
combine it with IEEE 874 (radix-independent floating point
handling) and otherwise *reduce* the number of choices open
to implementers.  There are various other goals but I think
those are the major ones.

Oh, one other thing in here that I applaud; they're standardizing a
hardware operation they call "fused multiply-add" which is key to
bigfloat support, and including an annex on working with arbitrary-
sized float formats.

				Bear