[news.eclipse.tools.uml2] Re: Composite attributes

[Ed Merks <merks@xxxxxxxxxx>]

Ricardo,

I don't know that much about UML so hopefully my comments are useful to 
the point...

Does Widget extend Element?  All your applied stereotyped objects will 
be modeled using dynamic EMF.  As such, if your profile model extends 
something that's generated, your dynamic instance will extend the 
implementation of that generated class.  But if your model doesn't 
extend something that's generated, it will just be an EObject and will 
be implemented by EObjectImpl, DynamicEObjectImpl, or anything else that 
behaves correctly.  You can use EObject.eClass() to determine the type 
of object you are dealing with and should find that the eClass() of your 
widget instance is the Widget EClass.  You should be able to do many 
cool and useful things using 
eObject.eClass().getEAllStructuralFeatures() to determine all the 
features of eObject and using eObject.eGet/eSet and so on to access or 
modify your data.  Using eObject.eAllContents() you can easily walk the 
entire tree of contained objects.


Ricardo Giacomin wrote:
> Hi,
>
> I have an stereotype Container with a property 'children: Widget[0..*]' 
> (iow, a list of widgets, where Widget is another stereotype and aggregation 
> type is Composite). When I create a model and apply stereotype container to 
> a given class, New Child menu options include a 'Children Widget' option. 
> Perfect, that is what I expected.
>
> My question is: I have a method that traverses the model and transform each 
> Element on it. However, the child of type Widget is not an Element but an 
> EObject (more precisely, a DynamicEObjectImpl). Isn't the widget I added as 
> child of my container an Element of the model? Is it possible to obtain the 
> Element from the EObject?
>
> By the way, I've read in a message on this group that it is not possible to 
> have an stereotype with a composition to another stereotype 
> (http://dev.eclipse.org/newslists/news.eclipse.tools.uml2/msg03179.html). 
> Does that mean what I did is not correct?
> Thanks a lot,
>
>