> I thought this whole discussion was based on a perceived need for > extra levels of type classification and fallback. perceived != demonstrated. I certainly haven't seen a convincing argument.