First my use case: I want to extend my EMF editor so that users can import instances of other ecore models, which differs in structure and naming. For example I've got ecoreA, which represents my EMF Editor. EcoreB should describe a part of EcoreA, but with differences in structure and naming.
So when a user wants now to import an instance of ecoreB and add it to ecoreA, the ecoreB instance has to be transformed so that it fit in for ecoreA.
I had a quick look around at the Epsilon Transformation language and wonderng if this would be a sufficient aproach for my use case or if it would be easier to provide my own parser?
Unfortunately I coudn't find any Tutorial describing how to use ETL. The examples or chapter 6 in the Epsilon book hasn't gave me enough informatins. So at the moment I'm not sure how to start with ETL.
I'd suggest starting by checking out and running the Tree2Graph example under . For a more complex example, please see . Once you've become familiar with the language you could have a look at  which demonstrates how to call ETL programs from Java. I hope this helps.