Re: Pizza Example

This is a very interesting discussion.

Codd developed normal forms in the context of the rest of the work he was doing. First normal form seemed like a good idea at the time. Making each normal form defined in terms of the prior normal form was just a mathemetician's idea of KISS ("reduce it to a problem already solved").

Perhaps what would be useful would be a theory of normalization that formalizes the way PICK developers model data, the way the relational normal forms formalize some of relational modeling. It doesn't have to start with the same axioms as relational normalization. And if it doesn't make sense to make each normal form prerequire the earlier forms, then don't.

Is there such a theory? Received on Wed Apr 07 2004 - 14:49:19 CEST