Re: Enforcing functional dependecy constraints

From: David Cressey <david.cressey_at_earthlink.net>
Date: Wed, 07 Dec 2005 14:34:04 GMT
Message-ID: <wpClf.507$nm.373@newsread2.news.atl.earthlink.net>

"x" <x_at_not-exists.org> wrote in message
news:dn1j2m$fpo$1_at_domitilla.aioe.org...
> Hi !
>
> Since this is a database theory group and I'm bored by the null and 3vl
> threads I ask yet another elementary question:
>
> Suppose you have this relation R(A,B,C) with the following functional
> dependencies AB->C and C->B.
> What is the best way to implement this in available SQL DBMS in your
opinion
> ?
>
> Regards,
> x
>

I'm not sure what you mean by "best". If it's my opinion of "what's best", I'm going to dodge the question by giving the universal answer: "It depends".

But before we move from relations to SQL, maybe we could discuss a little normalization:

You could decompose the relation R(A,B,C) into two relations S(A,C) and T(C,B).

Please bear with me for presenting relation T in a curious order. These two can be recombined, unless I miss my guess, back into R(A,B,C) without loss. I think S and T are better than R from the point of view of hermful redundancy and update anomalies. Is this relevant to your original question?

Do you have a different answer? Received on Wed Dec 07 2005 - 08:34:04 CST