The uml2's implementation is just one of the examples to which we applied
our reverse engineering algorithms. UML2 has many of its own rules to
generate the code based on the uml2 metamodel, so we think it deserves
deeper investigation.
I read your answers a couple of times, and will be looking at the code
generation package. I guess you are talking about the class
org.eclipse.uml2.codegen.ecore.genmodel.util.UML2GenModelUtil when
answering my 2nd and 3rd questions. But I only found
isFactoryMethods(GenModel genModel) and getKeyGenFeatures(GenClass
genClass) in uml2_1.1.1. I did found the ones you mentioned in uml2_2.0.1.
So should I check the latest implementation for how the code is generated?
Moreover, is UML2 code generator a separate code generator, and what's the
relationship between EMF code generator and UML2's? Any plans for the its
documentation?