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

Re: SRFI-77 with more than one flonum representation

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.

William D Clinger wrote:
If I understand you correctly,
you are asking for two operations that would round an arbitrary
inexact real to single or double precision, respectively, with
any underflow or overflow that accompanies the reduction in
dynamic range, and returns the result as an inexact real of
unspecified precision.

I agree that it would be good to add those operations to SRFI 77.
Did I state their semantics correctly?

Yes, except that I am not sure the the argument and result should be an "arbitrary inexact real". Other options are "flonum", which would mesh nicely with the efficient flonum part of SRFI 77, and "arbitrary real", which would be yet more general. For me, "flonum" for the argument and result would be most useful.

We need to be careful about single or double precision in this context and how these are related to s-exponent and d-exponent literals. That is, we need to make sure that it is clear that "round to single precision" refers to the hardware's idea of single precision and not to the Scheme's idea of the representation of s-exponent literals.


Dr Alan Watson
Centro de Radioastronomía y Astrofísica
Universidad Astronómico Nacional de México