RE: [wg@2uworks.org] Re: Book



Date view Thread view Subject view Author view Attachment view

From: Joaquin Miller (miller@joaquin.net)
Date: Tue 10 Sep 2002 - 14:35:12 BST


>I had thought (perhaps naively) that an object diagram would be a concrete
>syntax that mapped to instances of elements in the Semantic Domain, instead
>of the Abstract Syntax - but I can't work out whether that's what you're
>saying or not.

Let's keep at this discussion.

My ideas about what the setup should be are not based in practical 
experience with semantics, so are not really worthy of the attention of 
those who have experience.  But i feel i can state some requirements clearly.

1.  An architect must be able to declare specific individuals in a UML 
model, in order to specify the configuration of the system.

2.  A complete formal semantics must provide interpretations for these 
individuals.

........................

It is confusing at the very best (and i am too sweet to choose harsher 
words) not to have these objects in architects' models be in the same model 
and modeling language as the classes that describe the characteristics 
common to several of them.

It's puzzling (for untutored me) if part of a model is syntax and part of 
that same model is semantics domain.

We'll all benefit from those who do the hard work to provide formal 
semantics for our models, and to provide model checking tools, and model 
repositories, and metamodel tools to define language extensions or provide 
metadata to programs, and ...

But i strongly feel it will be best if any formal semantics comes as a 
whole that provides an interpretation for both class and object.

Date view Thread view Subject view Author view Attachment view