If you want the job done properly, this already exists: https://en.wikipedia.org/wiki/Decimal128_floating-point_form...