On the Casimir effect without cutoff