Very fine details when dealing with floating point numerics

See this for more about the binary floating point encoding and computation: https://en.wikipedia.org/wiki/IEEE_floating_point