---- On Thu, 29 Nov 2001, Eric D. Tarkington ([EMAIL PROTECTED]) wrote:
> > Re: (ROSE) RE: My BIG Question(long). > Les Munday <[EMAIL PROTECTED]> wrote: > > Answer: =93All class operations should be private to the class=20 > > that implements those operations, no class should be able to=20 > > directly access the operations of another class.=94 > > I have corresponded privately with Les about objects and how they model > events. He is proposing a radical change to our sense of how an object > works, and the only way to assess it would be to apply it for a few > months. Even that may not be conclusive, though, because existing tools > are not fully adapted to the revised object. > > As an old automation and communications guy from the bronze age, I am > sympathetic to much of what Les says, but it runs contrary to what I > think most people visualize as an object. > > In the conventional view, an event is modeled as a message, which is > modeled as an operation on some receiving object. Good coupling is > mainly achieved by a careful selection of what operations are public and > what private. If a message in a sequence diagram can be understood by a > pure domain expert, it is generally realized as a public operation (or, > at least, it is not private). > > Although we avoid functional decomposition, there is no harm in > perceiving that objects may be created to be shallowly or deeply nested, > and that shallowly-nested objects are usually more "architectural" (more > clearly related to the use case diagram, for instance). > > Although people make countless mistakes in applying this model, the > facts remain that it is very powerful and that a large number of people > have been able to gain an understanding of the model that is largely > shared. To the degree that the understanding is shared, we have culture > and industry. > > Les needs to demonstrate powerfully the benefits of his approach, which > may be superior to common practice. How he does this is a mystery to > me, but until he does it, the proposition falls to Tarkington's > principle: "It's better that it be standard than that it be good." > Eric, You give me too much credit. Because I posted the original question to many users groups, you won't have seen all the responses that I got to my original questions. In fact other people had spotted a similar flaw with the UML, and at least one company (small plug for Pathfinder here) claim to have come up with an add-in for Rose that helps alleviate my problem. > Tarkington's principle is irritating, and it is as much sociological and > psychological as technical, but I keep finding useful applications. > > At the level of practical suggestions, I think Les should develop a very > simple manual of how to apply his ideas to one or two popular > domains.=20 > He may also consider suggesting extensions to the UML, or drawing up an > altogether new language. > Eric, It's already done. At least the inital version. I'm stil collecting feedback. I have received a little from some readers, but not enough to convince me that I have explained my principles sufficiently. This URL was in the original email. http://home1.gte.net/res0hbt4/html/Process.htm Ok, I'm going to collate a summary of responses now, so don't disturb me for the next hour. Les. ________________________________________________ Get your own "800" number Voicemail, fax, email, and a lot more http://www.ureach.com/reg/tag ************************************************************************ * Rose Forum is a public venue for ideas and discussions. * For technical support, visit http://www.rational.com/support * * Admin.Subscription Requests: [EMAIL PROTECTED] * Archive of messages: http://www.rational.com/support/usergroups/rose/rose_forum.jsp * Other Requests: [EMAIL PROTECTED] * * To unsubscribe from the list, please send email * * To: [EMAIL PROTECTED] * Subject:<BLANK> * Body: unsubscribe rose_forum * *************************************************************************
