You can reasonably efficiently find the nearest rational number whose numerator and denominator fit in a given sized type. This means your rationals are imprec

You can reasonably efficiently find the nearest rational number whose numerator and denominator fit in a given sized type. This means your rationals are imprecise, just like floats; the sparse subset of numbers they can represent exactly is different from floats' (and possibly more useful).