[news.eclipse.modeling.mdt.uml2.ocl] Re: OCL and EMF 2.3 Generics

["Christian W. Damus" <cdamus@xxxxxxxxxx>]

Hi, Miguel,

Currently (and in M5), the OCL parser does not interpret the EEList data
type as equivalent of OCL's Sequence.  Rather, the collection type of a
multivalued feature is implied by the feature's uniqueness and orderedness,
as indicated in both the UML and OCL specifications:

  - multivalued feature of type T is ...
     - unique and ordered      ==> OrderedSet(T)
     - unique and unordered    ==> Set(T)
     - nonunique and ordered   ==> Sequence(T)
     - nonunique and unordered ==> Bag(T)

While Ecore does define an EEList data type, it does not define ESet,
EOrderedSet, EBag.

If you want to define an ETypedElement having a type that resolves (in OCL)
to Sequence(Sequence(CharacterObject)), then you would actually have to use
the OCL collection types in your model.  Otherwise, the type of your
*scalar* feature of type EEList<EEList<CharacterObject>> will be treated by
OCL as an opaque data type.  This probably isn't convenient.

Regarding your usage of the TypeResolverImpl:  this won't do what you want. 
Try this instead:

  private void displayOCLType(ETypedElement te, EClass ec) {
 
   EClassifier oclType = TypeUtil.getOCLType(te);
   String oclTypeName = getName(oclType);

HTH,

Christian

news.eclipse.org wrote:

> 
> Hi,
> 
> I've been looking at how OCL types an ETyped Element with an EGenericType,
> using this code:
> 
>  private void displayOCLType(ETypedElement te, EClass ec) {
> 
>   TypeResolver tr = new TypeResolverImpl(ec.eResource());
> 
>   EClassifier oclType = tr.resolve(te.getEType());
>   String oclTypeName = getName(oclType);
> 
> My question is, in 1.1M5 or later, what OCL types will be given to the
> following (for example):
> 
> a) an ETypedElement with eGenericType "EEList<EEList<CharacterObject>>"
> 
> b) In OCLTest.ecore (in the tests plugin, from CVS) the code above does
> not return
>    OrderedSet(Fruit), Set(Fruit), and so on, for the EReference
>    FruitUtil.orderedSet, and so on.
> Am I using the right APIs?
> 
> Again, this is not with the aim of opening a bug, but rather I would like
> to know the statement of direction for 1.1
> 
> Regards,
> 
> Miguel