This code is buggy because they are using floats internally. A 32bit float can not represent the exact measurements specified.

This is a perfect example of why just statically type checking your code does not make it correct.