Is it safe to define a NaN using 0.0/0.0? Why doesn't the C standard define a function to return a quiet NaN? (Or does it??? I don't think so.) What do some of the open source implementations of the standard library's math functions do? Plan 9 sqrt.c -- ach, where does NaN() come from?
Well, well, well: I should've tried apropos nan right away:
But I'm still a little unclear on the portability of these. Is it safer from a portability perspective to use strtod?
An article suggesting comparing floats in terms of the signed-magnitude numbers corresponding to the bit representations of the floats! Comparing floating point numbers. Cool!