Bug in example code for isnan in autoconf info

[David Cournapeau]

Hi,

    In section 5.5.1 (portability of C functions), there is a mention of
isnan and isinf. A portable replacement for isinf is suggested when not
available. Unfortunately, the function is wrong on two important aspects:

    - isnan(x-x) returns true for x being inf and Nan. But isinf(NaN) is
false
    - also, some compilers optimize the x-x to 0, this making it fail too.

To solve 1, I would suggest to define isnan and isfinite as basis, and
isinf as a compound of those two. The following definition for isfinite:

#define isfinite(x) \
    !isnan((x)-(x))

Should be ok.

For 2, I would suggest using ((x) + (-x)) instead of ((x)-(x)). I am not
really familiar with what IEEE-754 permits, but ((x) + (-x)) seems to
work much more often than ((x)-(x)) for compilers which optimize
aggressively.

regards,

Davud