>>How about loading them in as a string and then converting with VAL? I've done this sort of thing in the past using mainframe files.
>
>That solution is OK for ALL data types BUT the ones whish are represented in Floating Point and they are (for TP): real, single, double, extended & comp.
>
>To explain more closely:
>
>The Floating Point types store the binary representations of a sign, exponent & a significand. A represented number has the value
>
> exponent
>+/- significand x 2
>
>
>Some external library written in C/C++ as PLB & FLL (because I need it both for DOS/Win) is solution, but I can't find it. (Please don't tell me to write it, because I don't do C ;)

Hmmm. I used to program in TP & C, and I used these types, but never wrote binary to an output file, as it takes up extra space (back then space was important). Non-binary numbers the Foxpro VAL function should handle (float,real, etc.).

Have you tried searching MS KB for the info on the library you are looking for? (One problem in finding something like that is any C/TP prgmmr can write a converter pgm in a jiffy, so why create a library?) But perhaps there is one out there...maybe someone else knows of one. Sorry I can't help further...