Re: Bill of materials / Groups in Groups

Harry Chomsky wrote:
>
> joe_celko_at_my-deja.com wrote in message <85l2o8$8ri$1_at_nnrp1.deja.com>...
> >
> >Actually that is a cute phrase to define normalization -- the more
> >formal version is that inserts, updates and deletes do not cause
> >anomalys in the database. That is, the database is always in a valid
> >and internally consistent state at the end of a transaction. 1NF to
> >5NF, BCNF, DKNF, et al are _kinds_ of normal forms which prevent
> >certain kinds of anomalys.
>
> OK, you're using a definition of normalization that is quite different from
> any I've seen in a textbook. Essentially, given any database design and any
> set of business rules, you can make the DB be "normalized" (in this sense)
> with respect to the rules simply by adding constraints and triggers. So I'm
> not sure how useful your definition really is.

I would say that a constraint rejecting an input to the DB is an anomaly.

Kyle Lahnakoski                                  Arcavia Software Ltd.
(416) 892-7784                                        www.arcavia.com