On Fri, 22 Feb 2002, Ned Konz wrote:

> On Friday 22 February 2002 09:34 am, rallende at harlan.di.unc.edu.ar wrote:
>
> > What fo you think about the mapping wich speaks meyer, ie. class = module.
> [snip]
> > Although we still believe that class = modules is not a bad idea, it's not
> > closed topic for us, and any idea about it will help...
>
> Modules (IMO) should be the unit of packaging/re-use. As such, a single class
> isn't big enough. Often you have several classes doing internal work, and one
> or more classes presenting an interface to the rest of the world. The UML
> "class category" is a more useful concept, and it's closer to the Modules
> scheme in Squeak 3.3.
>
> Except that UML doesn't understand class extension (like DeltaModules).
>
> I had lunch with Jim Rumbaugh at OOPSLA '97 and asked him why not. His
> response was something like "we're all C++ guys and we never heard of being
> able to do such a thing". So much for a "Universal Modeling Language".
Sorry... but we did not use uml. For us a module is a "well defined
component of a software system, (and) we may view a module as a provider
of a computational resource of services" Carlo Ghezzi et.al. Software
engineering principles.

So, we have an artisan design (we did it following principles of these
books) and we need to conciliate (at least) this concept an equivalent
in o.o. software...

thanks
r.