inexact infinities

[alexshinn]

R6RS provides support for inexact infinities
and NaN objects. Do we keep these, and if so
do we use the same literal syntax and arithmetic
as in R6RS?

[cowan]

Testing Schemes with +inf.0 -inf.0 +nan.0, and if required (/ 1.0, 0.0) (/ 0.0 0.0):

PLT, Gauche, Gambit, Chicken[1], Bigloo, Guile, Chez, Chibi understand the syntaxes.

MIT, Scheme48/scsh, Kawa[2], SISC[3], SCM[4], Scheme 9, Scheme 7, VSCM do not.

SSCM doesn't seem to support anything but small exact integers.

[1] Chicken accepts the standard syntax, but outputs +inf, -inf, +nan without the .0
[2] Kawa uses the syntax #i1/0, #i-1/0, #i0/0
[3] SISC outputs infinity.0, -infinity.0, nan.0 but does not accept them on input
[4] SCM accepts and outputs +inf.0 and -inf.0, but uses 0/0 for NaNs?.

I recommend that we accept the R6RS syntax.

[kumoyuki]

Larceny certainly handles the +info.0 syntax. I am in favor of having explicit representations for infinities and NaN in any case...

[cowan]

The WG voted to accept this feature just as in R6RS.