Ben Pfaff <[EMAIL PROTECTED]> writes:
> * Is it acceptable for isfinite to raise an exception on
> a signaling NaN?
Yes.
> * Is it acceptable to #include <float.h> in and put
> inline functions in math.in.h?
Yes, to both. Of course 'inline' needs to be configured.
But do you really need float.h? See below.
> +static inline int gl_isfinitef (float x)
> +{
> + return x >= -FLT_MAX && x <= FLT_MAX;
> +}
How about 'return x - x == 0;' instead? That's easier to configure;
you don't need to worry about <float.h>.
