I agree with Ron that exporting parts of the model like we do when we
generate code would be a nice feature. Another nice improvement would be
to do the same with the import.

I'm facing similar challenges as Ron and what I did in the meantime y
merge the models when I generate the code.

To achieve this I create a dummy class on the specific models to
stablish the relationships and this dummy class is never generated. An
example is I have a module with the Party pattern classes in a separate
ArgoUML file so if I need the Party feature in a new project I create a
dummy Party class in the project ArgoUML file. In this way I can create
associations or generalizations to it. When I generate the code of the
project I leave this dummy Party class out because the real Party class
was already generated and coded using the Party ArgoUML. I don't know if
I was clear enough and hopefully this can help somebody.

Best wishes,
Diego

Ron Wheeler wrote:
> I did a quick test. The export seems to export too much (whole model).
> Perhaps this can be made to work but it would be nice to be able to
> select a package to export.
> I did not try the import to see how it deals with duplicates.
>
> Ron
>
> Mark Fortner wrote:
>> Oh I see what you're trying to do.  I misunderstood what you were
>> trying to accomplish.
>> If you have a core set of classes that you want to reuse in a
>> a number of different models you can simply export those core classes
>> into an XMI file
>> and import that file into other models.
>>
>> In your dependent projects, you just add those entities that you've
>> imported to the appropriate diagrams and add
>> whatever associations you want.
>>
>> I'm not sure if the -D parameter trick I talked about will keep
>> updating your dependent
>> diagrams, but that should be easy enough to test out.
>>
>> My earlier comments were around handling deployment of applications,
>> rather than construction
>> of applications.
>>
>> Hope I didn't confuse you too much,
>>
>> Mark
>>
>> On 5/25/07, *Ron Wheeler* <[EMAIL PROTECTED]
>> <mailto:[EMAIL PROTECTED]>> wrote:
>>
>>     It is a bit over my head but I can understand the direction that
>>     you are
>>     pointing.
>>     Your example is a bit different than what I am looking for but
>> perhaps
>>     not that different in solution.
>>
>>     I suppose that I can take the AndroMDA starting model, add my core
>>     classes (employee, business unit, etc.) and then save this as a
>> model
>>     for starting each portlet.
>>
>>     It appears that I still have the on-going problem of propagating
>>     changes
>>     made to the core classes to all of the models that might be affected
>>     when I go from one major version to another where most of the core
>>     classes get minor (I hope) enhancements.
>>
>>     Is there any solution for this?
>>
>>     Are there any tools for applying transformations to XMI?
>>
>>     Ron
>>
>>     Mark Fortner wrote:
>>     > Ron,
>>     > I usually create stereotypes and tagged values for the information
>>     > that I want associated with each object.  For example, if I
>> want to
>>     > create an EJB deployment descriptor, I take the fields that are
>>     > required for the descriptor, create a stereotype called EJB, add
>>     those
>>     > fields to the EJB stereotype, and export this as an XMI file.
>>     >
>>     > You can then import the XMI file or use it as part of a profile (a
>>     > collection of stereotypes) to suit your needs.  You can use the '
>>     > -Dargo.defaultModel=MyProfile.xmi' startup parameter to start
>>     ArgoUML with your profile.
>>     >
>>     > You can apply any of the stereotypes from the profile to the
>>     objects
>>     > in each of your projects.  When you're finished you can export
>> your
>>     > projects and XMI.  You have to apply a transform to the XMI to
>>     turn it
>>     > into a deployment descriptor, but that's about the limit of the
>>     work.
>>     >
>>     > I suspect that AndroMDA has the ability to handle those kinds of
>>     > transformations.  How you integrate this in to AndroMDA, is an
>>     > "exercise for the student", as my old Calculus professor would
>> say.
>>     >
>>     > If you don't want to rely on AndroMDA to handle the last bit,
>>     then you
>>     > can put this into an Ant target. It's simply a matter of
>>     unzipping the
>>     > ArgoUML, and applying your transform to the embedded XMI file.
>>     >
>>     > Hope this helps,
>>     >
>>     > Mark
>>     >
>>     > On 5/24/07, *Ron Wheeler* <[EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>
>>     > <mailto: [EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>>> wrote:
>>     >
>>     >     I am planning to use ArgoUMl with AmdroMDA to generate as
>>     much of a
>>     >     Jetspeed Portal as possible.
>>     >
>>     >     One of the issues that I think will come up in using
>>     Jetspeed is the
>>     >     need/ability to break down a fairly big application into a
>>     set of
>>     >     JSR-168 portlets.
>>     >     This implies, I think, that I am going to have a number of
>>     AndroMDA
>>     >     projects in ArgoUML ; one for each Portlet since each
>>     Portlet is an
>>     >     independent object that is loaded into the Jetspeed Portal
>>     >     depending on
>>     >     the user profile, etc.
>>     >
>>     >     I think that this is going to result in the same basic
>> entities
>>     >     (employee, business unit, code tables, etc.) appearing in
>>     many ArgoUML
>>     >     models.
>>     >
>>     >     It seems to make sense to define and maintain these entities
>>     in one
>>     >     place and import them  into each model in much the same way
>>     that the
>>     >     basic AndroMDA entities are incorporated into the model at the
>>     >     start of
>>     >     a AndroMDA project.
>>     >
>>     >     Is there some way to do this, using ArgoUML, so that
>> changes to
>>     >     the core
>>     >     entities can be easily replicated to all of the portlet
>>     models without
>>     >     having to go to each model and manually change each class in
>>     the core
>>     >     that has changed. The work is one issue; the probability of
>>     >     forgetting
>>     >     to make a change or making an error in making the changes in
>>     one
>>     >     portlet
>>     >     model is pretty high and will be hard to find when different
>>     portlets
>>     >     handle the same entity differently.
>>     >
>>     >     Suggestions about how the core entities should be organized
>>     to make
>>     >     portlet modelling easier would be appreciated.
>>     >
>>     >     Ideally I think that I would like to be able to export part
>> (or
>>     >     all) of
>>     >     a master model of classes and import them into other models
>>     >     overwriting
>>     >     any existing instances of the same entity.
>>     >
>>     >       
>> ---------------------------------------------------------------------
>>     >     To unsubscribe, e-mail: [EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>
>>     >     <mailto:[EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>>
>>     >     For additional commands, e-mail:
>>     [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
>>     >     <mailto:[EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>>
>>     >
>>     >
>>
>>    
>> ---------------------------------------------------------------------
>>     To unsubscribe, e-mail: [EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>
>>     For additional commands, e-mail: [EMAIL PROTECTED]
>>     <mailto:[EMAIL PROTECTED]>
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

-- 

**XNG **| Diego Gutierrez Zaldivar
Tel.: [54] 11 4780.0667

Fax: [1] 320 514.4271http://xng.dyndns.biz/

Reply via email to