[Modeling] Simplicity and design principles: ( Re: Seq. diag. answer to Michael Winikoff)

Dr. Hong Zhu hong.zhu1@btinternet.com
Tue, 25 Mar 2003 16:51:32 -0000


Renato and Michael are addressing a very important issue of design
principles, that is, simplicity. This principle is not listed in UML 2.0's
design principles. Personally, I would like to put this as one of the most
important design principles for AUML. The complexity of the specification of
a language is certainly related to misunderstanding and difficulty in the
implementation of tools.

Of course, there are two levels of simplicity. The meta-level is the
simplicity of the specification of the language. The model level is the
simplicity of the models, i.e. the language that enables modelling with
simplicity. AUML should be able to achieve simplicity at model level by
providing more meta-level concepts and mechanisms. But, how to achieve
simplicity at meta-level might be difficult.

-- Hong

----- Original Message -----
From: "Renato Levy" <rlevy@i-a-i.com>
To: <modeling@fipa.org>
Sent: Tuesday, March 25, 2003 3:55 PM
Subject: [Modeling] Re: Seq. diag. answer to Michael Winikoff


> Considering how forthcoming the CASE companies have been and how good (8^)
> ) are the tools, it will be a fight to get to support any AUML at all. I
do
> agraa that were we can simplify UML 2.0 we should try to. I find it a very
> "bulky" for the lack of better word notation.
>
> ps: Michael, I also personally agree that the icons would work better. I'm
> just playing devil's advocate, because I know that we will be questioned
in it.
>
>
> >Dear Michael,
> >
> > > Briefly, I believe that some sort of suggestive icon is a lot easier
to
> >read and
> > > process than a three letter acronym (especially for casual users of
the
> >notation
> > > - "what does ign mean again?")
> >
> >Sure! It is better also for non English speaking people. But UML 2.0 has
> >already defined acronyms and all UML users will know them. That's a pity
> >that we are late little bit.
> >
> > > Regarding multiple notations, any half-decent tool should be able to
> > > automatically switch presentations, i.e. show "alt" as "?" (or
whatever
> >symbol
> > > is chosen) so I don't see this as a significant issue.
> >
> >Yes, but we should not have any expectations on dynamic behavior of CASE
> >tools supporting AUML. I do not think that the key-players in the area of
> >UML 2.0 compliant CASE tools will implement also AUML with optional
> >notations. At least not in several months (years?) after AUML
specification
> >releasing.
> >
> > > Ideally, I'd argue for REPLACING the three letter acronyms with
> >symbols/icons,
> > > however it is too late for this, so I'm just suggesting that we define
and
> > > permit an alternative presentation. I believe that this will AID the
> >readability
> > > and understandability of the notation. I appreciate your concerns over
> >different
> > > notations, but in this case the difference is very small and, more
> >importantly,
> > > completely localised.
> >
> >OK We can offer presentation options in AUML for these elements, in sense
of
> >presentation options occurred in UML specification.
> >
> >BTW looking at UML 2.0, I feel that there could be more cases for which
we
> >will offer alternative notations. Not only replacing some acronyms by
icons,
> >but also replacing maybe several modeling elements by one visual element
(of
> >course with unambiguous metamodel mapping). E.g. while modeling some AUML
> >concept that is naturally represented by a dependency and a constraint.
It
> >can be drawn e.g. as a line with special arrowhead and guard, instead of
> >drawing dependency and constraint as two separate visual elements...
> >
> >Regards,
> >
> >Rado1.
>
>
>
>
> Renato Levy
> Principal Scientist
> Intelligent Automation, Inc.
> 7519 Standish Place, ste 200
> Rockville, MD 20855
> phone: (301)294-5241
> fax: (301)294-5201
> WWW: http://www.i-a-i.com
>
>
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org
> http://fipa.org/mailman/listinfo/modeling