"systems" programming languages where you care about things like fixed size integers, should have ranged number types and they should force you to deal with potential overflows explicitly when you're going from a larger range to a shorter range

e.g. if you try to add two 32 bit integers (with no explicit range) and store the result in a 32 bit integer, you will need to deal with the overflow. but if your initial 32 bit integers had an explicit range, the result might not need an overflow check.