Re: 2NF Controversy

Nomicon <NoSpam_at_hatespam.net> wrote:

>1) One side suggests that in order for a table to violate 2NF, the table
>must have attributes that are dependent on only part of a compound
>PK--this implies that 2NF violations only apply to situations in which a
>table has a compound PK. This requirement has been stated explicitly in
>several definitions.

     Yes, that is about it. I note a slight problem in that if a column did not depend on the key at all, to me, the table would not be in 2NF.

>2) The other side suggests that tables violate 2NF when non-key
>attributes apply to multiple rows. This definition also applies to
>tables with simple PK's.

     No. It is entirely possible that many records could contain the same value for one column. For example, a mailing list could have 1,000 people living in Vancouver.

>Which defintion is correct? Or are neither correct?

     The first.

Sincerely,

Gene Wirchenko

Computerese Irregular Verb Conjugation:

     I have preferences.
     You have biases.
     He/She has prejudices.