modf is for doubles, no?

[Benjamin Kosnik]

> My Stroustrup book says that modf() is for doubles, not floats. 

see 26.5 C library, p 6. The signature in the standard is 

float modf*float, float*)

> I assume the test is correct. What is being tested in this file?

Probably some of the autoconf stuff not being completely trashed. There's
a bunch of configuring that goes into bits/std_cmath.h, and it's a test to
make sure some of the more obvious things are not completely broken.

-benjamin