Re: Logical equivalence of simple and complex types under the relational model?

"Alfredo Novoa" <alfredo_at_ncs.es> wrote in message news:41aded45.6070921_at_news.wanadoo.es...
> On Tue, 30 Nov 2004 21:48:26 +0100, "Rene de Visser"
> <Rene_de_Visser_at_hotmail.com> wrote:
>
> >I am not sure that the split between complex and simple is well defined
at
> >all.
>
> It is not well defined at all.
>
> For instance one type might have several representations. Some of them
> may have one component and the others may have several.
>
> For instance we may have a date value represented by an integer
> (741543) or represented by a char array ('2004/12/2') or represented
> by an array of integers (2002, 10, 25)
>
> Regards

And am I correct in thinking that if we have a relation containing a date e.g. ("Sally", '2004/12/2')
we can create an updatable view corresponding to this relation where the tuple looks like:
("Sally", 2004, 12, 2)?

Or is there some catch I'm missing?

It also seems to me that we can consider '2004/12/2' as a atomic from Codds definitions and consider
the 2004, 12, 2 to be properties (as defined by Codd) of '2004/12/2'. Or am I missing something here?

Which makes it difficult for me to understand Codd' s statement

"A domain is simple if all its components are atomic (nondecomposable by the database system."

and the requirement that all domains be simple,

When this requirement does not seem amount to anything even within one model.

Or was he trying to get at something completely different?

Rene. Received on Wed Dec 01 2004 - 17:35:33 CET