Re: Normalization and DBMS

"x" <x-false_at_yahoo.com> wrote in message news:40a0f206$1_at_post.usenet.com...
> **** Post for FREE via your newsreader at post.usenet.com ****
>
> Codd 1970 ACM paper:
> "A first-order predicate calculus suffices if the collection of relations
is
> in normal form."
> "Such a language would provide a YARDSTICK of linguistic power for all
other
> proposed data languages."
>
> Someone claimed in this group that normalization is not "good" .
> How powerful is the data language of an (existing) DBMS not based on
> "normalization" ?

It is 1NF with which I disagree and all other normal forms are based on this. 2nd and 3rd normal forms (for example) are semantic and related to "functional dependencies" and they make a lot of sense for designing for minimizing the costs of the database over time as requirements change. 1NF is not based on semantics, but on a "guess" that making our mathematics simplest wrt to the language of the propositions/predicates would make the most sense. That is the only logic I can find behind 1NF and I have found absolutely NO and I mean NO emperical data to suggest that putting data into 1NF helps minimize the cost of the system for the long haul. In fact, the anecdotal evidence that I have seen (which could very well be skewed as I have done no scientific survey) is heavily skewed toward databases that are not in 1NF.

If you use Date's terminology, you need GROUP and UNGROUP operators in your language. This could also be NEST and UNNEST. The language used with PICK, for example, reads easier than SQL and includes such language as "WITH EVERY" so you get

LIST STUDENTS WITH EVERY MAJOR <> "MATH"

Did that answer the question? --dawn Received on Tue May 11 2004 - 22:09:12 CEST