Re: Status of UML 2.0



Date view Thread view Subject view Author view Attachment view

From: Andy Evans (andye@cs.york.ac.uk)
Date: Sat 26 Oct 2002 - 09:39:56 BST


----- Original Message -----
From: "Bernhard Rumpe" <Bernhard.Rumpe@in.tum.de>
To: <kokar@coe.neu.edu>
Cc: <puml-list@cs.york.ac.uk>
Sent: Saturday, October 26, 2002 7:13 AM
Subject: Re: Status of UML 2.0


>
> *****  FOR THE ATTENTION OF THE MAILING LIST OWNER  *****
>
> See X-Diagnostic: headers above for more details
>
> *****  FOR THE ATTENTION OF THE MAILING LIST OWNER  *****
>
>
>
> > I think a UML model defines constaraints on the programs that are consistent with the model. And
constratints don't
> > need to be ambiguous just because they are constraints. They describe subsets or classes of >
things. Is a<b
> > ambiguous on say natural numbers?
>
> Generally following Mitch's argument, I would like to add that I think
> we can
> distinguish two types of "abmbiguity":
>
> - "ambiguity of a statement"
> - "ambiguity of the language itself"
>
> - "ambiguity of a statement" could also be called "imprecisesness of a
> statement":
> 0 < x < 1000
>   is less precise about x than   99 < x < 102.
>   But the language used is perfectly precise.
>
>
> - "ambiguity of the language" or "impreciseness of the language" arises,
> if the
>   same language construct has (slightly) different interpretations by
> different
>   readers / contexts. E.g.
> "x is about 100"
>   is imprecise, because we cannot clearly identify the borders.
>
>
> --> With ambiguous languages it is difficult to make unambiguous
> statements.
>
> --> Modelling is largely about making imprecise (= not detailed,
> underspecified)
>     statements using unambigous languages, which in principle works.
>
> I think that UML suffers from being not defined precisely enough, but
> practicioners
> also make advantage from this fact when they want to specify
> ambiguously.
>
>
> And of course there is no principle problem to give a language such as
> the UML
> a precise semantics (This has been done for petri-nets, DFDs, etc long
> ago).
> The problems are:
> - UML is large, so it's a management problem
> - UML has many stakeholders. If we have a formal semantics that cannot
> be
>   generally agreed, what is the benefit?
>   Deriving a formally defined ad generally agreed semantics is not a
>   technical, but a social problem and may be largely unsolvable.
>
> Bernhard
>
>
>
>

Date view Thread view Subject view Author view Attachment view