On Wed, Apr 30, 2014 at 04:32:54PM -0400, Susan Hares wrote:
> Juergen: 
> 
> I suggest the UML as an Informational model is quick and easy to read.  Most
> generated prototypes can enforce interoperability you wish for.  If not, the
> UML to code tool is just broken.  The steps are: 
> 
> UML--> Data Model (yang/forces) --> code    

I believe that a tool that generates a meaningful YANG data model out
of a UML information model requires so many details in the UML model
that the UML model stops being an information model. A data model has
to be very clear about naming. If you use YANG, you need to transfer
an arbitrary graph of classes and their relationships into a tree.
The answer to the question what becomes a reference between branches
in a tree and what can be captured through nesting does not naturally
fall out of a UML class diagram. Perhaps I am overly pessimistic.

A key point of an information model is that it focusses on fundamental
concepts and that it on purpose leaves out details that are needed in
data models. So either your tool transforming UML class diagrams into
YANG has a second information source or you have to overload your UML
diagrams with details that turn your UML diagrams into a data model.
 
> I think the UML is readable. Please comment on my UML drawings that I sent
> at the beginning of this post.  If you wish a power point of the UML so you
> can edit it, I'll send one.  
> At the Yang step, Andy assures me it is
> readable so debugging the tool should be useful.  I was answering the
> performance issue question with the iterative code.  I think this does
> provide what you require. 

I like to see the UML input and the YANG output or even better the
tool. I am happy to see the text I wrote above proven wrong.

> Can you tell me where your experience states this is a misstep?

See above. I have seen people who added lots of details to their UML
class diagrams in order to drive automation until the UML diagrams
filled walls and essentially lost their value of summarizing key
concepts.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: +49 421 200 3587         Campus Ring 1, 28759 Bremen, Germany
Fax:   +49 421 200 3103         <http://www.jacobs-university.de/>

_______________________________________________
i2rs mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/i2rs

Reply via email to