Re: functional dependencies

(This is andrewst writing under another guise, due to problems with DBForums today)

rp_hathiwala_at_yahoo.com (Rohan Hathiwala) wrote in message news:<3ec1cded.0306112248.709008b9_at_posting.google.com>...
> So does the whole process of normalizing relations comes down
> to identifying FD's where the attribute that functionally
> determines the other one is non-prime in the present relation and
> then normalizing it so that it (the attribute) becomes a
> primay key in the new relation?

Yes

> If so then I believe
> we will have FD's only in theory but we will never ever have FD's
> in practice.

Only if you don't consider functional dependence on the key to be functional dependence, which you don't apparently (due to the definition of an FD that you quote). Yet clearly, there IS a functional dependence of all non-key attributes on the key in a normalised relation: given the employee number, there is only one possible value for the employee name, i.e. employee_name = f(employee_number). The fact that you'll never see the dependency stated twice in a relation is one of the goals of normalisation: elimination of redundancy. Received on Thu Jun 12 2003 - 11:47:37 CEST