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

From: David Cressey <david_at_dcressey.com>
Date: Mon, 23 Jul 2001 11:35:36 GMT
Message-ID: <c_T67.28$YQ.1535_at_petpeeve.ziplink.net>


> Even better case for trying to understand how a completely OO
> based application would be designed. Lee is making the point
> about the many needs for users to get at the data. How is this
> accomplished in OO only systems?

I think this is close to the root of the matter. Essentially, what the pure OO people are saying, if I can put words in their mouth, is that hiding the data from the users is a service. If they have full access to the objects, and if the object model covers the subject matter, then why should they need access to the data, except for data made available to them by the objects that they can access.

I'm skeptical of the above claim, but not completely closed to it.

If the only access is through the objects, then issues surrounding SQL all go away. Issues about multithreading, concurrent update and ACID don't go away, but they get rephrased in OO terms. There remains, IMO, an issue about the data model that describes the data exchange between the users and the objects. This is a consequence of the object model in the application, but if it's completely chaotic from one application to the next, then I'll stick with the SQL model for data exchange for the time being.

I know, I know, the SQL model isn't the relational model. But it's close. And, if people define their data well in some mainstream product like, say Oracle, the Data Dictionary will describe the data to some level of detail, to the user who knows the Data Dictionary, but doesn't know the application. I don't know what the OO equivalent of that might be.

--
Regards,
    David Cressey
    www.dcressey.com
Received on Mon Jul 23 2001 - 13:35:36 CEST

Original text of this message