One of the things related to MMTF that I’m not happy about is the way it represents relationships between models: the relator models can be pretty huge. That’s because the related models are represented in their abstract syntax as collections of EMF objects. If you add to that the “meta-muddle” that is the UML metamodel, even simple models can blow up considerably when represented in abstract syntax.
This post is mostly to serve as a reminder to myself, when I get some free time to look into metamodel pruning. This is a very nice paper: ”Meta-model Pruning”, by Sen et al.