Re: Alternative method to normilize tabels ...

From: Jan.Hidders <hidders_at_hcoss.uia.ac.be>
Date: 23 Oct 2002 10:02:58 +0200
Message-ID: <3db657b2$1_at_news.uia.ac.be>


In article <3ylt9.174224$8o4.28075_at_afrodite.telenet-ops.be>, joske_at_home <vandenbergeDOTj_at_village.uunet.be> wrote:
>
>If I apply these normalisation (2NF to 3NF) rules to my tables ... I have
>attributes that depend on 'a part' of the key AND a 'non key' attribute.
>I think that I need 'move' these attributes to a new table that get as key
>the 'part key' AND the non-key attribute of the original table ...
>
>Pff ... trying to express in another form ...
>
>Orig table:
>
>ORIG ( pk X, pk Y, Z, data, ...) attribute 'data' depends on X AND Z
>
>After split ...
>
>ORIG (fpkX, pkY)
>NEW (pkX,pkZ, data)
>
>Does it make sence?

Almost; you should also leave the column Z in the original relation. The general rule is that if you split off a dependency X1,..Xn -> Y1,..,Ym then 1. in the original table you remove Y1,..,Ym 2. you create a new table with columns X1,..,Xn,Y1,..Ym

  • Jan Hidders
Received on Wed Oct 23 2002 - 10:02:58 CEST

Original text of this message