64 bits is certainly the solution I would go for as well. Arbitrary precision numbers are going to involve heap allocation which is something I would normally

64 bits is certainly the solution I would go for as well. Arbitrary precision numbers are going to involve heap allocation which is something I would normally like to avoid entirely in a system like this.