I think the lack of it (project composition) in the Executable UML
book was intentional. (more reasoning below)
> Short final comments:
> * A domain (component + interfaces) is just one the domains under
> consideration in a project.
This should be qualified with, "by a number of teams" or "at different
times". A domain is supposed to be able to stand alone. Bridges are
anonymous couplings, so the contents of other domains are irrelevant
with respect to the modeling of the domain under consideration of the
modeler(s).
> * The Domain Subsystem mainly covers the modelling elements of a leaf-
> component in a larger-scale system.
... in tool X. Tool Y might have something completely different for
project composition.
> * Since Project and Domain/Component are separate subject matters, the
> Project metaclass should be placed in a separate Project Subsystem.
> * Since we must preserve PIM vs PSM separation for cross-platform
> reuse, a "model compiler" is a separate subject matter too.
> * How Domains/Components are "linked" (= wired across interfaces) must
> be modelled and, thus, covered by separate metamodel subsystems.
OK. I'm saying separate domain, which correlates with separate subject
matter.
In any metamodel part like the miUML Domain Subsystem the fact that
there is a concept called "project" has no relevance.
(no relevance to the subject matter(s) under development/maintained/
being updated)
How different tools, real projects or companies handles the concept of
"projects" is unimportant wrt the model(s) under development.
One can just hope that the tool manages projects in a proper way.
--
You received this message because you are subscribed to the Google Groups "miUML" group.
To post to this group, send email to star...@googlegroups.com.
To unsubscribe from this group, send email to starruml+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/starruml?hl=en.