Re: So let me get this right: (Was: NFNF vs 1NF ...)

David Cressey Said:
> I don't quite agree. Here's an excerpt from the on-line copy of [1]:

Well, I agree with everything you've just said (except for what I think your definition of "complex domains" is).

I would clarify matters by asserting that a system possesses RVAs when relational operators can be applied to values in a relation's attributes. I *think* Codd's term "nonsimple domains" refers to precisely this kind of situation. For example, complex numbers are a fine domain in mathematical logic, and are 'non-simple' in the sense of being 'non-atomic'. Complex numbers--as well as rational numbers, vectors, and matrices, etc--are not excluded so long as their contents cannot be accessed using project/restrict/join etc. Just data in relations, thank you very much.

For 'non-simple' domains where the contents of the domain *can* be accessed with project/restrict/join, Codd introduces "normalization".

(Confession: I have not read D&D's latest thoughts on RVAs, so I might be mis-characterizing their position.) Received on Thu Feb 10 2005 - 23:10:29 CET