The article assumes the use of a signed integer, by passing in values that can't be represented with a signed integer you can't assume the same solutions hold.