hi luis.
Luis Sergio Oliveira schrieb:
[...]
yes, ArgoUML will apply. See the thread
http://argouml.tigris.org/servlets/ReadMsg?list=dev&msgNo=21853
and page http://argouml.tigris.org/googlessoc2008.html
thanks for that, didn't see it before.
I'm not offering myself as a mentor, since I already committed to a
different project. But your idea seams very interesting. With a MOF
based extension mechanism ArgoUML could be used to build meta-models,
which would facilitate extra-UML modeling. There were some extra-UML
projects, being one example the PetriNets ArgoUML module
http://argopno.tigris.org/.
Do you think that this kind of project would be easier by using your
idea?
sure. i know some petrines theory, but i don't know exactly how they
should model real-time-systems. anyway, this kind of project would
definitly benefit from the things i have in mind.
afaik statemachines can be treated as some kind of subset of petrinets,
maybe it's possible to reverse this. a mof-description could derive the
things of the normal uml's statemachines(or activity graphs) and set
some constraints to it(like a PetriNetConnection may only accept
PetriNetStates as endpoints). this would be a specialization of existing
uml-stuff for a new purpose.
there are two ways to achieve this. both methods need argouml to define
a metamodel in mof.
the first way is to generate the code-stubs for the new types. amof2
does this and then need implementations for getting/setting the new
attributes, etc. the problem is, that argouml has to use methods it
doesn't know, so there is probably more to do.
the second way is a "pure reflective" approach. this means, that a newly
defined metamodel can be used straight away, argouml would explore it
using the reflective methods that mof defines and offers the appropriate
gui-elements to create and modify instances of the metamodel's types.
things might getting hard for both methods if the created elements want
a special visual appearance. since i don't even know, how argouml draws
the existing uml-stuff i can't tell if this is possible, what's needed, etc.
i've read a little about amof2, so i'm sure it can do the needed things.
there was in question about being "pure reflective" in their mailing
list, the answer was something like "normally it's not, but you can
easily achieve this" and a rough guide how to.
i don't know about euml, it's abilities and how mof comes in there.
i just wanted to start by using the uml 1.4 and therefor the mof 1.4
facilities, maybe it's possible do head towards their 2.x specs.
Or what other practical purpose do you have in mind?
i want do build the python-rte-modul, so i have a "modular" version of
it in mind, that can plug in special handlers for custom, mof-defined types.
Regards,
Luís
yours
alexander
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
