This Tuesday 2nd of November, the Eclipse Symposium opened the first afternoon of the ESE 2010 [1]. In this post we will give you an sightsee of the emergent projects presented during the session. The EMF symposium is mainly focus around demos of new projects and initiatives in the modeling world. Going further it gives an overview of the important projects and the next directions of the modeling projects.
Trends
The main topics this year were around Xtext, EMF repository, live UI model and model migration. Xtext was clearly shown as an important project and many initiatives around it were presented. Xtext [2] is a DSL for creating meta-model and text-based editor, it was shows in project such as
- Eclipse help: a DSL for generating Eclipse help documents
- Syntactic view: able to dynamically show the syntactic representation of the defined grammar
- JDT Integration: showing that the JDT has been integrated through an adapter layer to XText. As a result, the Xtext meta model can directly refer to Java classes contained in the project class path
The EMF repository was the second major trend. Today, we can see a real need in term of model storage which must be differentiated from traditional SCM. Then, we saw CDO [3] evolving to disconnected world in order to be used as repository. A more suited project was presented to help SVN to store models. It is based on EMF Store [4] and MTF. This combination enables to share a project with SVN and sharing models with EMF store. Therefore, a commit will trigger an SVN commit following by a EMF Store commit. This kind of tools are mandatory in projects which blends source code and models.
Last year there was few presentations about live UI model such as Wazaabi [5], but the initiatives were limited. Today we can see more project going toward this direction. This is the case of the RedView [6] project from Riena called the emf dynamic view. This shows clearly the direction concerning dynamic and next generation UI.
The next topic is the direct consequence of having used models for few years: the model migration. This is something really trendy this year, and really needed by companies working in modeling. If you use a model respecting a meta-model in the V1.0 of your application, it is likely that 4 years after the first version, the V4.0 would need a new version of the meta-model and to migrate existing models. This is not a trivial job, we need to manage particular cases, containment/uncontainment relationships, cardinality, etc. The COPE project [7] have been developed for this purpose. It uses EMF compare for comparing the two meta-models (considered as models of the same meta-model, ECore in that case) and from the diff model it proposes a set of actions to correct differences. As a result we have an history model tracing all the actions to apply, from which we can generate a migrator plug-in.
Finally we have to cite the Papyrus [8] project which is the most important UML project in the community. Coming from industry, Papyrus has been released in open source as an Eclipse project. It is mainly a complete UML tool but it also contains plug-in for industrial UML specifications such as MARTE, SysML, CCM. This project is going to become the de-facto standard framework for UML design.
References
[1] EMF Symposium, www.eclipsecon.org/summiteurope2010/
[2] Xtext, www.eclipse.org/Xtext/
[3] CDO, http://wiki.eclipse.org/CDO
[4] EMF Store, http://www.eclipse.org/proposals/emf-store/
[5] Wazaabi, http://wazaabi.org
[6] RedView blog by Riena, http://redview.wordpress.com/howto/examples/example-from-scratch/
[7] The COPE project, http://wiki.eclipse.org/COPE
[8] Papyrus, http://www.papyrusuml.org