[Fwd: [Modeling] Comments on interaction diagram modeling
Wed, 19 Mar 2003 17:47:35 -0500
I second Nodine, IF roles are correctly defined then one should be able to
describe a protocol in a open system based only on role playing.
At 10:26 AM 3/19/2003 -0600, you wrote:
>Whoops, I meant to send this to the entire list. Attached is my response
>Received: from thumper.research.telcordia.com (thumper [22.214.171.124])
> by austin.research.telcordia.com (8.9.3/8.9.3) with ESMTP id KAA10696
> for <nodine@austin>; Wed, 19 Mar 2003 10:17:52 -0600 (CST)
>Received: from austin.research.telcordia.com (austin [126.96.36.199])
> by thumper.research.telcordia.com (8.12.8/8.12.1) with ESMTP id
> Wed, 19 Mar 2003 11:17:43 -0500 (EST)
>Received: from research.telcordia.com (goliad [188.8.131.52])
> by austin.research.telcordia.com (8.9.3/8.9.3) with ESMTP id
> Wed, 19 Mar 2003 10:17:41 -0600 (CST)
>Date: Wed, 19 Mar 2003 10:17:41 -0600
>From: Marian Nodine <email@example.com>
>Organization: Telcordia Technologies
>X-Mailer: Mozilla 4.79 [en] (X11; U; SunOS 5.8 sun4u)
>To: "Wagner, G.R." <G.R.Wagner@tm.tue.nl>
>Subject: Re: [Modeling] Comments on interaction diagram modeling doc
>X-Virus-Scanned: by AMaViS - amavis-milter (http://www.amavis.org/)
>"Wagner, G.R." wrote:
> > > The UML 2.0 format is effectively <agentName:agentRole>, where you can
> > Why not keep the format <agentName:agentRole/agentType>?
> > This would allow to express Lin's problem of modeling
> > the interaction for shop assistants by using lifelines
> > associated with agentrole/agenttype combinations, like
> > <:customerWelcoming:shopAssistant>.
> > -Gerd
> > ---------------------------------------
> > Gerd Wagner
> > http://tmitwww.tm.tue.nl/staff/gwagner/
> > Dep. Information & Technology
> > Eindhoven University of Technology
> > Email: G.Wagner@tm.tue.nl
> > Phone: (+31 40) 247 26 17
> > Fax: (+31 40) 247 26 12
> > _______________________________________________
> > Modeling mailing list
> > Modeling@www.fipa.org
> > http://fipa.org/mailman/listinfo/modeling
>Where Lin started out (as I understood it) was with the comment that agent
>interaction protocols more intuitively map to interactions between roles as
>opposed to interactions between agents. This is related directly to the
>late binding issues of agents, I think. AUML is most challenged in the
>areas where agents have little resemblance to objects, as that is where the
>distinctions between AUML and UML2 will be the clearest.
>I do not really see the point in putting a specific named agent into the
>lifeline header, as this defeats the late binding approach. Perhaps we
>could represent the name with respect to the instantiation of the role, in
>the tab in the upper left-hand corner of the UML frame?
>I do see what Lin says with respect to roles possibly being specified in
>the context of an agent type. However, this should not be true in an open
>agent system -- due again to the dynamic nature of such systems and the
>late binding issues, no agent types are necessarily known. It seems to me
>that someone should come up with a good and concise definition of "role"
>that we can all agree on, and then discuss how to set it into the context
>of AUML interactions.
Intelligent Automation, Inc.
7519 Standish Place, ste 200
Rockville, MD 20855