RE: Status of UML 2.0



Date view Thread view Subject view Author view Attachment view

From: Bauer, Robert (rbauer@rational.com)
Date: Thu 24 Oct 2002 - 18:53:02 BST


Heinrich,

But my point is that your goal,

My understanding of "unambiguous" UML is that each UML diagram has well
defined meaning as opposed 
to several meanings depending on who is looking at the diagram. This does
(and should not) exclude 
the possibility that the same meaning can be expressed by different
diagrams.

is quite difficult, if not impossible to achieve.

For example, given some program P, I wish to determine it's meaning.  To do
so requires that I have
a formal semantics, one that is unambiguous and complete in the sense that
every sentence constructed
in the implementation language has only one interpretation in the
specification language - but that requires
me to show that the specification language itself is unambiguous - which in
general can't be done.

robert

-----Original Message-----
From: Hubert Baumeister [mailto:baumeist@informatik.uni-muenchen.de]
Sent: Thursday, October 24, 2002 10:20 AM
To: puml-list@cs.york.ac.uk
Subject: Re: Status of UML 2.0


Hi Robert,

Bauer, Robert wrote:
> Heinrich,
> 
> Although I don't read German well, I was able to look at the web pages 
> you referenced below.  One thing bothers me.  As I looked at the 
> www.2uworks.org website I noticed the goal of being "unambiguous" uml.  
> This does not seem to be possible unless uml was no more expressive than 
> a finite automata - that is, one could represent uml using regular 
> expression (or equivalently a left/right linear grammar).  Now, I 
> haven't thought about this very long or deeply, but it seems to me that 
> while it is relatively easy to check whether two uml "graphs" are 
> equivalent at the syntactical level, is intractable at best and quite 
> possibly impossible - if we were to use 1st order logic to express the 
> semantics, well at best it is only semi-deciable whether two 1st order 
> expressions are equivalent.  Since, ocl cannot yet express all of the 
> meta-model, it cannot possibly express the semantics of every uml model, 
> so while one may be able to decide whether two ocl expression are 
> equivalent (though I suspect even though ocl is not a 1st order logic, 
> it is still undecidable due to the presence of "allInstances"), the 
> "rest" of the semantics of the uml model are either in a natural 
> language (clearly impossible to determine in general whether two 
> sentences are equivalent or equivalently whether a sentence is 
> ambiguous) or anything more expressive than a F.A. (also except for 
> discrete context free languages, clearly impossible to determine 
> equivalence, ambiguity, etc.).
> 
> So, for those reasons, I am not sure why the goal is stated as it is.

My understanding of "unambiguous" UML is that each UML diagram has well
defined meaning as opposed 
to several meanings depending on who is looking at the diagram. This does
(and should not) exclude 
the possibility that the same meaning can be expressed by different
diagrams.

Greetings,
	Hubert

-- 
Dr. Hubert Baumeister, Institut für Informatik, Universität München
mailto:baumeist@informatik.uni-muenchen.de
http://www.informatik.uni-muenchen.de/~baumeist
phone (x49-89)2180-9375  * fax -9175




To remove yourself from this list please mail
puml-list-request@cs.york.ac.uk
with a message containing the word "unsubscribe".

Date view Thread view Subject view Author view Attachment view