Re: Normalization by Composing, not just Decomposing

From: Jan Hidders <jan.hidders_at_REMOVETHIS.pandora.be>
Date: Tue, 13 Apr 2004 20:56:39 GMT
Message-ID: <buYec.70606$NG1.4810792_at_phobos.telenet-ops.be>


Dawn M. Wolthuis wrote:
>
> Would it be conceptually easier to talk that way too?
> I would like a pizza with pepperoni on it
> I would like a pizza with mozarella cheese on it
> I would like a pizza with feta cheese on it
> I would like a pizza with a tomato-based sauce on it
> I would like a pizza with a Chicago-style crust
>
> There, I've got my talkin' normalized! We should teach children to talk
> this way since it is conceptually somuch easier, eh?

That's not what the relational model claims. The idea is that if you have multiple application that look at the data in different ways, then it could be that some application would like to see the data nested like this:

   { (ordered-pizza, { (ingredients) }) }

and another one might want

   { (ingredient, { (ordered-pizza) } }

In that case it is easier and more practical to formulate the central conceptual data model in a flat way, so that it does not favour any of the applications inadvertently, and also does not favour a specific physical implementation inadvertently. Data-independence and all that.

In theory, anyway. ;-)

  • Jan Hidders
Received on Tue Apr 13 2004 - 22:56:39 CEST

Original text of this message