> But, I agree that if the input to removeUnused satisfies the exhaustive > and irredundant invariants, then a comparision of the number of used cases > with the number of used datatype variants will be sufficient to determine > if the default case is needed. Agreed.