Re: A Normalization Question
Date: Wed, 21 Jul 2004 02:00:43 GMT
Message-ID: <f7kLc.129451$a24.12695_at_attbi_s03>
"Neo" <neo55592_at_hotmail.com> wrote in message news:4b45d3ad.0407201715.7a04b317_at_posting.google.com...
> > > A data model that decides which type of redundancy is good or not is a
> > > limited model.
> >
> > but using redundant references to a single location storing
> > "brown" is not redundant becuase the redundant references
> > are metadata, which is okay. Therfore, your own model
> > decides which types of redundancy are good and which are not.
>
> In general, references don't need to be normalized.
Why does your model need references? RM doesn't even have references. Since, as Date argues, references are neither necessary nor sufficient, I take the presence of references in a model to be a sign of unnecessary complexity.
Marshall Received on Wed Jul 21 2004 - 04:00:43 CEST