Re: Primary-key ... alternate key ... (newbie question)

From: Jan.Hidders <>
Date: 4 Oct 2002 15:21:39 +0200
Message-ID: <3d9d95e3$>

In article <sC0n9.144230$>, joske_at_home <> wrote:
>In many normalisation articles, they sometimes mention to mark the alternate
>Why is it important to 'mark' these? Where are they used during the
>If you choose for a certain combination of attributes as a primary key, does
>the other attributes, (that are part of the alternate key), influence the
>normalisation process?

Yes, they do. And the texts that suggest that you only need to look at the primary key are simply wrong. In fact, the notion of primary key has no place in normalization theory, at all. You just talk about the set of candidate keys. For example, if you want to know if you are in 3NF you should check that for every non-trivial dependency that ends in a non-key attribute (i.e., an attribute not in any candidate key) the determinant is the superset of a candidate key. The words "primary key" are nowhere to be found in that definition, as you can see.

  • Jan Hidders>
Received on Fri Oct 04 2002 - 15:21:39 CEST

Original text of this message