On Fri, 2011-12-02 at 19:01 -0800, Andrew Lentvorski wrote:

I think that expansion is wrong.  IIRC, IEEE754 specifies that NaNs are 
unordered.  Consequently, I think that both (nan1 == nan2) and (nan1 != 
nan2) must return false.

That's true in C, and in Gambit

(= nan1 nan2) => #f

because it conforms to IEEE floating-point equality.

But eqv? compares floating-point numbers bitwise in Gambit, and I agree with that behavior.

Brad