Re: Clean Object Class Design -- What is it?

From: Rico <TrooperRico_at_bugplanet.com>
Date: Sat, 21 Jul 2001 23:34:24 GMT
Message-ID: <%gG17.946$G42.1113_at_newsfeed.slurp.net>


"Bob Badour" <bbadour_at_golden.net> wrote in message news:1pF17.4171
> >
> >There's no advantage to be gained by infiltrating your application with
 the
> >specifics of your database system. In fact, modern software engineering
> >practices would frown on this kind of thing. He's right to separate the
> >data storage model from his application. RDBMS, OODBMS, or otherwise.
>
> True. He should insulate each application's view of the data from the base
> relations using views. He is doing none of this which cripples the
 database.
> He is taking his objects and mapping them directly to physical structures
 in
> the database foregoing both physical independence and logical
 independence.
>

A view is a derived relation. As you're aware from reading The Third Manifesto, we don't really treat them any different from a base relation. Which means you don't pollute your application with RDBMS-specific material like this when it's possible to avoid it.

>
> >I know what normalization is. Now, care to answer his question? Give a
> >specific example of these "esoteric update anomalies." I've got
 ObjectStore
> >server on my research system, let's watch it die.
>
> Any design with mult-valued dependencies or join dependencies that is not
 in
> 5NF will do.
>

Let's have a specific case. Stop talking generalities. Received on Sun Jul 22 2001 - 01:34:24 CEST

Original text of this message