Re: 2NF There are two Definitions which is right

"Jon Heggland" <heggland_at_idi.ntnu.no> wrote in message news:MPG.1d49eebc5b69db69896fb_at_news.ntnu.no...
> In article <3k7t18Ft6e3bU1_at_individual.net>, jens_at_haase.to says...
>> I found two definitions for 2NF:
>>
>> 1: A relation R(A,F) is in 2NF, when every attribute not belonging to
>> the primary key of R is fully functionally dependent on the primary key
>> of R
>>
>> 2: A relation schema R is in 2NF if every nonprime attribute A in R is
>> fully functionally dependent on the primary key of R
> [8<]
>> Which definition is right?
>
> The second. Normalisation theory doesn't deal with the arbitrary primary
> key distinction. The first definition probably makes the assumption that
> there is only one key.

While the conclusion might or might not be correct, it is not necessarily true because of the single primary key assumption. BCNF, an aspect of Normalization theory, deals specifically with cases where multiple arbitrary candidate keys exist satisfying certain conditions. You are familiar with Date, so see chapter 11, section 11.5 for example.

I believe both definitions are correct if total context is given. Typically authors make a single primary key assumption when introducing normalization for pedagogical reasons.

> --
> Jon

• Dan