Re: [xtm-wg] TM Conceptual Model: Semantics of the UML "modern dinosaur"?

[Murray Altheim <murray.altheim@eng.sun.com>]

"W. Eliot Kimber" wrote:
> Robert Barta wrote:
[...]
> > Well, UML certainly helps to understand relationships between items
> > under specification but - I agree - UML itself is notoriously
> > "meaningless".
> 
> It is the responsibility of the modeler to define the intended semantics
> of the UML graphical components they use. This is done both through
> prose documentation of the model itself and through the use of
> stereotypes that specialize the syntactic components. We were certainly
> careful to do this in the initial modeling work we did in Paris.
> 
> One cannot expect to be able to read a UML diagram in isolation any more
> than you can a DTD or source code--it must be supported with additional
> documentation that clearly explains the intended semantics.

This has certainly been my criticism of our current use of UML, ie., that
while you were gracious enough to provide a 'how-to' document, 

  a. I'm skeptical that our current use of UML in the spec truly conforms
     to this early definition, or that this could even be demonstrated.
  b. absent this additional documentation actually appearing in the 
     specification, the UML in the spec is essentially meaningless, well,
     essentially definition-less at least. This seems dangerous.
  c. our current "UML" seems to rely to a great degree on the embedded 
     textual notes to make clear its concepts, which to me rather avoids
     using UML itself to clearly and concisely create the model. IOW, if
     the UML doesn't suffice to describe the model, why not just use text?
     The idea of a software tool "interpreting" the diagrams would obviously
     ignore the textual notes. A textual description could at least be as 
     clear as the UML diagrams, and not create an abstraction that can be
     misinterpreted by people holding different understandings of the UML 
     graphical symbols.

A model is only as good as peoples' understanding of the modelling 
language, and when a modelling language is basically a general purpose
toolkit that can be modified to fit an application, it's hard to 
understand how it is suitable for pedagogical use. If a map of New York
used light blue lines for the roadways, some people might mistake them
for rivers and streams. No legend?

I think UML has certainly provided great benefit within the CMS in 
providing a framework on which they could come to agreement on a model
(although I still don't see that agreement in practice, since there 
still seems to be such disagreement on some of the fundamentals), but 
I remain to be convinced that putting it in the spec is a good idea*,
especially absent the documentation describing our specific use of UML.
I think it may end up being a real source of confusion, especially to
those who think they understand the UML but really don't.

I was initially quite enthusiastic about UML, especially with the idea
that we'd have that three-level model (as described in Paris), and that
we could actually use software tools to generate our DTD. Given that 
this to a great degree turned out to be a fantasy (either due to the
limitations of the modelling language, the software, our time 
constraints, it doesn't matter), I'm much more skeptical about its 
real efficacy.

Murray

*This even given the trouble I went through to provide "pretty" versions
of the UML diagrams for the spec.
...........................................................................
Murray Altheim, SGML/XML Grease Monkey     <mailto:altheim&#64;eng.sun.com>
XML Technology Center
Sun Microsystems, 1601 Willow Rd., MS UMPK17-102, Menlo Park, CA 94025

      In the evening
      The rice leaves in the garden
      Rustle in the autumn wind
      That blows through my reed hut.  -- Minamoto no Tsunenobu

To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com