Casimir force between two ideal-conductor walls revisited