Re: Multi Valued Interface Models?

JOG wrote:
> Hi dawn, two levels of quotes can quickly career away from the main
> point (for instance I don't believe it is all just a case of
> nomenclature but I think we would get very bogged down along that
> path). The following lines stood out though:
>
> >> Now if you are defining a new meaning for "data model" that corresponds
> >> to an interface that is fine
> > No, I am contending that is what Codd's def of data model is about. It
> > is all about the representation of the data at the interface point
> > between a client of a database and a database service.
>
> I believe many would contest this, and think Codd's intention was the
> exact opposite - that his definition of a data model is all about
> getting away from representation,

Then what does the information principle mean? "The entire information content of a relational database is represented in one and only one way: namely, as attribute values within tuples within relations." (Date, Edgar F. Codd, A Tribute, www.sigmod.org/codd-tribute.html)

This sure sounds like representation to me. In addition, I don't see how it could be about anything other than representation, but this representation could be in various places. In the case of the information principle, it seems to be in the interface between any client and the database.

> because information, with its complex
> internal interrelationships, may have a multitude of possible but
> equally valid representations.

agreed

> A data model surely aims to eliminate
> bias,

I don't think so

> structuring data and providing mechanisms to extract it out in
> any way we see fit for the task or information requirement in hand.

But the data model biases will show through in the costs incurred in getting from one place to another

> The blog often speaks of a UI data model and that the RM does not
> always "fit" it. I'd contest that a UI can "have" a data model in Codds
> sense of the term at all.

I will agree that Codd was specifically talking about data models related to "large shared data banks" but that does not mean that the same concept of a data model cannot be anywhere else (apologies for the double negative).

> Is a UI not rather a reflection of an
> underlying data model,

No. If you come up with a workable definition of a data model as in "relational data model", you can see that you can identify data models throughout software. That is not what Codd did, however, as his focus was the database.

> like a fire casting a shadow on a wall?

This shows evidence of a database-centric view of the world. A database is but one aspect of software development. Yes, there is a data model associated with the interface to the database, but there is also a data model associated with other interfaces.

> Manipulate the fire and one can get a different static shadow - the
> former is doing the work, and the latter it just an artefact.
> (Apologies for the cheesy analogy.)

It showed up a difference in our perspectives, so it was useful. Cheers! --dawn. Received on Tue Feb 14 2006 - 05:25:38 CET