Single precision versions of libm functions

[Ian Lance Taylor]

My employer is using uClibc in a project, although I am not much
involved with it myself.  We recently discovered that the version we
are using does not implement the single precision versions of the math
functions (sinf, etc.).

I see that the current SVN sources have a float_wrapper.c which does
the math in double precision, but that is not ideal for us, since our
processor implements single precision (32-bit) floating point
arithmetic but not double precision (64-bit) arithmetic.

As it happens, I implemented the single precision versions of the
float routines for the SunPRO library back in 1994, for newlib.  My
implementation has been picked up and patched by NetBSD and glibc,
among others.

I'm writing this note because I'm wondering if there is any reason why
uClibc did not pick up those single precision implementations.

Thanks.

Ian