Soft 32 bit floating point implementation, a pretty abnormal one. There are plenty soft floating point libraries in the wild, I am not satisfied with either of those. All of them are way too complex.
This is a tiny project, the goal is to make a simple and understandable implementation of single-point floating point precision numbers. I do not care about efficiency as long as the code is simple. I'll write a complete series of articles covering all bits of the code.