[r6rs-discuss] division by exact 0
agj at alum.mit.edu
Fri Jun 22 10:30:11 EDT 2007
| From: William D Clinger <will at ccs.neu.edu>
| Date: Thu, 21 Jun 2007 06:28:15 -0400
| The specification of / says the divisor(s) "must not
| be (exact) 0", but goes on to provide these examples:
| (/ 1.0 0) => +inf.0
| (/ 0.0 0) => +nan.0
The restriction about "(exact) 0" was not in R5.93RS. I can find no
discussion of this on r6rs-discuss, nor any formal comment addressing
this issue. The editors should not have made this change without
public notification or discussion.
Aren't we supposed to be closing on a draft for ratification? The
editors should be removing stuff, not adding it at this late stage.
There are many who do not hold the view that exact 0 is "stronger"
than 0.0. It is unreasonable to forbid an implementation from
coercing an exact 0 argument to inexact in a mixed exactness call.
More information about the r6rs-discuss