Re: Normalization and Derived Information

>> I'm wondering about the precise status of derived columns in terms
of 1nf,
2nf and so forth. <<

Tom Johnston did a series of articles on other forms of redundancy than normalization problems that would be worth pulling up.

It is bad if you materialize the computed column because it can get out of synch with the columns used to compute it. We used to have that problem all the time in punch card systems.

But technically the table can be normalized up to Domain Key Normal Form (DKNF), where we are allowed to have contraints like " CHECK (total_price = qty_ordered * unit_price)" as part of the definition as well as FDs. Received on Thu Sep 23 2004 - 04:47:44 CEST