Meetings

Skype meeting 28/08/2012

• Progress presented: server emf generation fixed. Client emf model query stabilized.
• Found bug in emf model registration. Wrong emf ecore registration extension point.
• Extra query proposed by Istvan: find remote references (what dependencies will be new in the next commit).
• Footnote for implementation: interesting question would be to discover code metrics (with graph properties like spanning tree, longest path, largest connected components, inflexion point, etc).

Skype meeting 07/08/2012

• Presented the improvements I have done: licensing, selectable projects, more reliable emf model building.
• I should clarify what is a product: a single project in the workspace or an Eclipse product.
• I should take into account, how to generalize the solution over eclipse plugins, not just single-jar projects.
• Presented the 5 types of the discovered dependencies (Class usage, method call, field access, inheritance, method override).
• Istvan clarified the development process for the IncQuery:
  • Create example.javamodel file -> export it through my view -> load it in the editor -> load it to the development engine.
• It is possible to have two import declarations in the query file.
• For the next meeting (21th of August):
• Learning IncQuery usage.
• Create queries for my model.
• Connect the "big" and the "small" workspace model in the queries.
• Optimize the production of the workspace Emf model.
• screenshot
• Future work:
  • How to use IncQuery as a library.

Skype meeting 03/07/2012

Beginning:

• Presented the progress of my work (Dependency discovery works but really slow and introduced some bugs which has to be eliminated in order to proceed).
• Presented how to obtain Java elements through the EMF model (JavaCore.create() with the handlerIdentifier string).

New goals:

• Relatively fast model building in the modelfinder project.
• Serialized EMF model of the workspace state and the dependency graph built my previous solution.
• Get the slice of the dependency graph model (drop unrelated parts) and shrink it to a minimal size.
• Main question: how much do we gain with this elimination and how much false positive results do we introduce in the result set.

Further work:

• The same as last time, if the models are ready to evaluate we should import them into the IncQuery system end execute queries upon it.

Important part of IncQuery understood:

• We can utilize queries upon multiple models with different metamodels by joining the two with the query languages (much like an sql join operation).

Skype meeting 19/06/2012

Beginning:

• Presented current state of work (workspace->EMF model, dependency analysis infrastructure on the uni's vm, wiki).

Licensing:

• Source headers: just author name should present (for now).
• Look after the technology transfer topics.

For the next meeting:

• Required feature: identify workspace elements based on the EMF objects.
• Required feature: put dependency relations into the model based on JDT informations.
• If it is ready, ask Bence to help initiating queries on this model.

Further work:

• Find a way, to merge the two EMF models: the one which represents the workspace state and the other, which comes from the database.
• Need to select a subset of the model to load on the client part (we want to transfer just a really small model to the client side).

Other:

• Have a look on EMF-CDO, whether it is usable for our solution. If we used it, it would let us work with EMF models directly on the server side and on the client side too.
• If we have time/energy, we can do a comparison between Neo4j and CDO.