Re: two nasty schemata, union types and surrogate keys

From: Brian <>
Date: Sun, 27 Sep 2009 07:36:51 -0700 (PDT)
Message-ID: <>

On Sep 27, 9:57 am, paul c <> wrote:
> Brian wrote:
> ...
> > And it's wrong!  Normalization has everything to do with the semantics
> > of the data.  If you look carefully at all of the examples of
> > converting from one normal form to the next higher form, you'll see
> > that the schema before conversion and the one after conversion are NOT
> > equivalent in their capacity to express facts.  Instead, the schema
> > after conversion has at least the same capacity to express facts, but
> > not exactly the same capacity.  For example, a 2NF schema that is not
> > in 3NF, ...
> This could be extremely misleading to the casual reader.  In one breath,
> it suggests a set-piece that recasts relations under the guise of
> normalization but ignoring constraints at the same time.  In fact, when
> normalizing, one must nearly always introduce constraints in order to
> preserve "semantics".  Most trade writers make the same mistake which
> may be why so many people think re-design involves only normalization.

I don't think it is misleading. It does not suggest anything: it states fact. The example supports that fact. In fact, the 3NF scheme in the example includes an inclusion dependency, a necessary consequence of the transitive dependency in the 2NF schema. Received on Sun Sep 27 2009 - 16:36:51 CEST

Original text of this message