Re: foundations of relational theory? - some references for the truly starving

From: Bob Badour <bbadour_at_golden.net>
Date: Mon, 3 Nov 2003 17:22:56 -0500
Message-ID: <CtOdnSqF9973SDuiRVn-gw@golden.net>

"Dave Best" <davebest_at_usa.net> wrote in message

news:ea757642.0311031305.1beca7a4_at_posting.google.com...
> "Anthony W. Youngman" <thewolery_at_nospam.demon.co.uk> wrote in message
news:<fHrE46GYsFo$EwWN_at_thewolery.demon.co.uk>...

> > In article <CUYmb.86063$Ms2.64480_at_fed1read03>, daveb
> > <davebest_at_SuPsAaM.net> writes
> > >"Ross Ferris" <ross_at_stamina.com.au> wrote in message
> > >news:26f6cd63.0310260541.7a6a9af9_at_posting.google.com...
> > >> "Bob Badour" <bbadour_at_golden.net> wrote in message
> > news:<GumdnaAjFvrJmQaiU-KYvg_at_golden.net>...
> > >> > The values in a foreign key reference are redundant because they
appear
> > in
> > >> > multiple relations. In this case, the redundancy is appropriate and
> > >> > necessary to represent the data.
> > >>
> > >> Interesting "admission", or at least an observation. Of course this
> > >> redundancy is ONLY necessary because of the "flat earth" nature of
SQL
> > >> implementations.
> > >>
> > >> If the data were stored in a multi-valued database, or even an XML
> > >> data store, then the redundant data could be removed.
> > >
> > >No, you have merely encoded the redundancy in the structural
relationship.
> > >
> > Where? There is no key (foreign or otherwise) with which to do the link,
> > because there is no need to do a link.

> >

> > So yes there is a structural relationship, but there is no redundancy
> > because no information is stored - it is IMplicit in the data store, not
> > EXplicit.

> >

> > Cheers,
> > Wol

>
> The foreign key is in the logical model.  Any nested relation has a
> primary key consisting of the primary key columns of the containing
> relation (and which form a foreign key to it) plus its own primary key
> columns.  In Pick, the primary key of a MV element is the primary key
> of the record plus its array index.
>
> A relational database with relation-value (nested) attributes can
> choose to physically store them clustered with the containing
> relation, and so does not have to physically store the redundant
> columns.
>
> The logical model of the data can be thought of as the API which a
> program uses to access the data.  This is distinct from the way it is
> physically stored on the disk, which can be anything the vendor
> chooses, including the Pick method of value-encoded variable-length
> strings (so long as this representation is not exposed by the API).

Dave,

You have to stop and consider to whom you are replying. Wol is ignorant and stupid.

For instance, the relational model requires the dbms to represent all information as explicit data values in relations. Because of this, one can manipulate any information using the same simple, powerful tool: predicate logic. As soon as one represents information in any other way, one must either accept limited functionality or increased complexity or both. In Pick's case, it is clearly both less functional and more complex.

As evidence of Wol's ignorance and stupidity, he actually thinks that implicit physical representations of information have some advantage. He lacks the cognitive ability to comprehend that the redundancy still exists at the logical level, and the only thing different is Pick has less functionality for greater complexity.

How many informed and intelligent people want to pay more and receive less? Can you imagine someone walking into a store and declaring he wants to pay top dollar for pure crap? What would you conclude about such an individual? Received on Mon Nov 03 2003 - 16:22:56 CST