Re: Nested Relations / RVAs / NFNF

From: Marshall Spight <mspight_at_dnai.com>
Date: Wed, 27 Oct 2004 17:43:26 GMT
Message-ID: <17Rfd.316248$MQ5.24677_at_attbi_s52>


"Kenneth Downs" <firstinit.lastname_at_lastnameplusfam.net> wrote in message news:lkkolc.9oe.ln_at_mercury.downsfam.net...
> erk wrote:
> >
> > The key is a set of attributes, not the value of those attributes. The
> > "key-ness" of an individual attribute (its membership in one or more of
> > the candidate keys) is orthogonal to its type.
>
> I think what Marshall is heading for is based on the observation that a
> surrogate key allows certain operations and disallows others. We commonly
> use integers for surrogate keys, but these integers allow other operations
> that make no sense for keys and can cause bad data. So if we want to
> maintain the tradition of strong typing and the use of surrogate keys, then
> Marshall's idea is a step forward on solid ground.

Thanks!

I note that in the type theory world, attention is paid to the appropriateness of associated operations. For example, one might choose to have separate *types* for metric units and for English units, and allow the type system to either prevent invalid crossovers or convert for you. Also, you can do things like have a separate numeric type for speed, distance, and time, or even things like length, area, and volume. The type of the multipy operation for (length, length) is area; for (length, area) is volume, etc.

Marshall Received on Wed Oct 27 2004 - 19:43:26 CEST

Original text of this message