Double precision floats can losslessly manipulate integers with up to 56 bits precision, so you get the best of both worlds with no extra effort.