R: [Modeling] Modeling tools

Massimo Cossentino cossentino@pa.icar.cnr.it
Fri, 27 Jun 2003 17:35:54 +0200

Dear all,

I'm doing the same work with the methodology TC. I consider important
two different requirements for our tool: 1) the possibility of creating
a SPEM and AUML profile; 2) the availability of an API (possibly
supporting Java) in order to allow the introduction of specific
meta-methodology capabilities (for example a support in the methodology
composition). Some of these tools, have been already discussed in other
mails of the modeling TC. Sorry for repeating but I have a little
different point of view (the API requirement).

I already contacted some companies with these results:

1) Rational (IBM). Rose has reached its limits. They are moving towards
a new line of products (XDE). Their API allows reading the attributes of
the project elements but not writing in them (by now at least)

2) MetaEdit. Already a CAME/CASE tool (the only one in my knowledge).
This is an advantage for the methodology TC. It is possible to (easily)
create the profiles we need. An API will be available within one month
but I'm not sure if it will support Java.

3) Visio. Not really a CASE tool but more a graphical tool. Very easy to
produce a SPEM/AUML profile. We already have a SPEM stencil available
from our site. The integration of the UML module in it, is proprietary
and it is not possible (easy at least) to read and write the project
properties (about it, I also had some interesting but useless phone
calls with a programmer of the MS developing team). I had some
experiences in programming Visio (with VBA). Good but only for little

4) EclipseUML. Very interesting also because of the perfect integration
with the development environment (Eclipse). (BTW I have some doubts
about the real intention of IBM of continuing the development of
Eclipse, rumors...). An API will be available shortly (it seems...).
Being a Java tool I suppose the API will be Java too. 

5) Eclipse Universal Modeler. It seems interesting. By IBM. A member of
my lab is in touch with the developing staff. By now the project is
sleeping but they are available to discuss about a new impulse. I think
this is a very interesting opportunity. I'll keep you updated.

6) ArgoUML. Never liked it. I tried to explore the possibility of
expanding it (it's open source) but the code is not well documented at
all (at least one-two years ago)

7) TNI-Valiosys. It has been used in ADELFE. I've not experienced it
directly, but it seems that it has not a good support for profiling and
it lacks of an UML diagram (I don't remember if activity or sequence).

8) MagicDraw. I like it. I have a reasonable hope but I'm not totally
sure about the profiling possibilities, I'm further exploring. The API
is well documented and complete.  I'm starting to work with it.

The cost problem. I'm not sure if it will be possible to obtain what I'm
going to propose, but I think that we could ask to the selected tool
company some kind of special low-cost (or even free) license for our
standardization work. It will be a good thing for a company to have its
tool adopted by a FIPA TC and probably we could succeed.

I think that adopting an unique tool for both the methodology and
modeling TCs could be a natural choice. The methodology TC will adopt
the next FIPA modeling language in describing the method fragments, the
MAS meta-model and so on. Obviously a FIPA goal is that future MASs will
be developed both in AUML and following the FIPA meta-methodology
approach. We could try to converge towards an unique choice in order to
facilitate the language-methodology integration, if it will be possible.


Massimo Cossentino

> -----Messaggio originale-----
> Da: modeling-admin@fipa.org [mailto:modeling-admin@fipa.org] Per conto

> di James Odell
> Inviato: marted́ 24 giugno 2003 14.23
> A: ModelingTC
> Oggetto: [Modeling] Modeling tools
> Hi All,
> > Are we going to use a common tool, like Rational Rose?
> You mean IBM Rose?  :-)
> Seriously, this is something that we all need to begin pursuing now.  
> We need to find tool companies that  will support Agent UML: either by

> their modification or ours.  (An example of a tool that we can modify 
> would be
> Visio.)  In order for Agent UML to be successful, we must
> have a choice of tools, otherwise people will not be inclined 
> to adopt it.
> So, the question for this TC is: how can we get Agent UML modeling 
> tools? Let's start making a list, and I will post it on the website.  
> IMO, this is very important.
> -Jim
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org http://fipa.org/mailman/listinfo/modeling
> ---
> Incoming mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.470 / Virus Database: 268 - Release Date: 08/04/2003

Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.491 / Virus Database: 290 - Release Date: 18/06/2003