On 2020-10-25 10:40 -0400, John Cowan wrote:
> However, I'm thinking that it would be better to write a direct C
> implementation and then provide a sample of FFI glue required to get the
> information in and out of Scheme.  That way we escape all dependency on
> bytevectors and deal only with floats, doubles, and ints across the
> Scheme/C boundary.  I don't know if you are up for doing that.

I'm not sure I want to commit to writing something like that, since I
can't immediately visualize how much work it would be.  How much of a
Scheme number system should we assume in order to write a good example
of this concept?  Would it be sufficient to provide C versions of
nan-quiet?, nan-negative?, etc., operating on doubles?  This would be
fairly easy to wrap, I suppose, although my only FFI experience has
been with Chibi.

--
Wolfgang Corcoran-Mathe  <xxxxxx@sigwinch.xyz>

"Abstraction lies at the heart of all programming languages."
--Greg Michaelson