On Sep 16, 2009, at 12:25 PM, Ed Merks wrote:
How about?
This problem was addressed by the GMT UMLX project. A Model Registry was
developed that was independent of any particular modeling project. This
solution migrated to the M2M QVT Declarative project. A further
migration is necessary to allow the MDT OCL project to provide the
OCL editor currently forming part of the M2M QVT Declarative project.
A few weeks ago I asked if it would make sense to contain this technology in the OCL project because from what I gather, much of this solve problems very specific to OCL and QVT, i.e., trying to come up with a name/identifier for each package rather than using it's namespace URI.
I agree.
IMHO the OCL-specific name mapping belongs to the OCL project.
The EMF Index can provide the indexing of EPackages in the workspace (i.e. find the corresponding resource by the EPackage's nsURI).
Contribution to EMF/Index was investigated, but EMF/Index is about smart use of known models
rather than location of models.
EMF Index is primarly about locating models by their name.
This is a pretty vague statement that even experts would have trouble turning into a meaningful interpretation and it's this type of vagueness that's made it difficult to figure out exactly what you're proposing and why you consider it to be a generally reusable component. There is for example
https://bugs.eclipse.org/bugs/show_bug.cgi?id=220218
which sounds similar on the surface but the design I expect to see for that sounds very little like what you're proposing.
I think that the EMF Index solves the mentioned bugzilla by making it possible to look up EPackages by their nsURI.
The aliasing or shortening of the nsURIs belongs to the different clients (e.g. OCL).
