I'm trying to build a 32-bit float out of its 4 composite bytes. Is there a better (or more portable) way …