Re: On view updating

Costin Cozianu <c_cozianu_at_hotmail.com> wrote in message news:<2r6lvdFl18ghU1_at_uni-berlin.de>...

> I think they've got it wrong on several counts.
>
> 1) Is whether view updating is crucial to "logical data independence".
>
> Very little have I seen in the way of defending this point.

Because it is trivially true. What is not crucial is the automatic view translation. We can always hardcode the updating rules procedurally in triggers.

> Since
> relvars are "variables" and the rest are computed values,

Views are variables too. Virtual variables but variables. Although a "closed view" has nothing of virtual to the user.

> 2) That the principle of orthogonal design has any pragmatic meaning,
> and should have a decissive contribution to shaping the relational model.

Darwen rejects the Principle of Orthogonal Design, and Date is reconsidering it, if it was not dropped yet.

IMO The POD is a failed attempt to patch the strange behavior caused for not using the rules of logic in the translations.

If we change the union update criterion:

P(A) or P(B)

by:

P(A) xor P(B)

Then the strange behavior dissapears even if we violate the POOD.

> Now by mandating the view update sucessful resolution, we want to
> transform relational model into Prolog.

In something better than Prolog IMO.

> We don't need that because if we want Prolog, we already have it.

As you know, Prolog has many limitations.

I don't see anything evil in having a powerful inference engine. It may save a lot of tedious and error prone procedural coding.

Regards Received on Tue Sep 21 2004 - 00:44:11 CEST