William D Clinger
will at ccs.neu.edu
Wed Aug 30 16:27:38 EDT 2006
> I don't have any more time to waste on this debate right now; suffice it
> to say I don't believe the R5RS says what you think it says, and I think
> the R6RS requirement is overly restrictive for systems that represent
> inexact numbers using IEEE arithmetic. Nevertheless, the R6RS wording,
> which I missed (or missed the implications of) on my earlier readings,
> exists, and I'm not going to ask that it be changed at this point. Given
> that it exists, and that it requires (/ 0 3.5) => 0.0 (or at least some
> other inexact number), it doesn't make sense for (/ 0 0.0) to return 0.
> So please go ahead and change the (/ 0 0.0) example to return +nan.0. You
> might also add an example showing that (/ 0 3.5) evaluates to 0.0.
Thank you. I will make those changes.
> Mike seems to think you want to change the (* 0 +inf.0) and (* 0 +nan.0)
> examples too:
That wasn't my interpretation of what he wrote, but in
any case I do not want to change those examples, because
I think they are correct in the current draft.
More information about the R6RS