Re: views of binary operations
Date: Mon, 17 Jul 2006 17:51:03 GMT
Bob Badour wrote:
> Data independence comprises two distinct types of independence. Physical
> independence allows one to adjust performance characteristics without
> disturbing a given logical design. Logical independence allows one to
> adjust the logical design without disturbing existing applications. One
> achieves the latter form of data independence through application views
> -- especially updatable views.
> Physical independence gives independence between the base relations and
> the physical storage. Logical independence gives independence between
> the base relations and the relations that applications interact with.
> Data independence is a relative concept and a principle both for
> designing dbmses and for evaluating the quality of dbmses. Thus, while
> one could create a product using relations and the relational algebra
> but lacking named views, one would be hard-pressed to establish that the
> product meets the requirements for a dbms.
> I am not sure exactly when the name 'view' was introduced, but it was
> certainly already well-established when Codd published his '12 rules'.
Thanks, that second paragraph seems to explain the distinction rather well. It's what I would have called application independence which was why I also wondered how it was fundamental - it's fundamental, I guess, not so much to RT but rather in the sense that in practice it's a fundamental technique. In fact, I vaguely recall that some measure of appl'n independence was possible with the hierarchical or network products before Codd came along with his 'symmetrical exploitation'. For me, the really fundamental technique at play in views is indirection, but no matter, I've no serious argument with what you say.
What troubles me more are the logical arguments against the updating of certain views, especially the unpredictability of deleting from certain joins. The big names like McGoveran, Darwen and Date seem to be somewhat at odds over this and apparently there is further explanation in the latest edition of 'TTM' which I don't have a copy of, but I found Darwen's earlier arguments pretty convincing in layman's terms.
p Received on Mon Jul 17 2006 - 19:51:03 CEST