Section: Overall Objectives
Highlights of the Year
Year 2009 has seen the consolidification of the scientific and technical approaches proposed by the AtlanMod team. This is notably visible in the three following points:
The choice to base most of the operations on models on rule-based and metamodel-based declarative expressions has become mainstream. We published the first paper proposing this in 1998 at the EDOC conference in San Diego  . Today this idea has gone mainstream and most of the competing teams in the domain are following similar graph-based approaches. The advantages of using rule-based declarative formalisms is accepted not only in academic research, but in industry as well. This approach, proposed in the original paper by R. Lemesle, has been shown to be the best one for facilitating the maintenance of modeling applications. The partnership with the OBEO company has allowed the long research process on transformation languages to converge to industrialization and broad dissemination. As a consequence, the AtlanMod team may now refocus on the ATL-Research initiative.
A second option that was taken by the team is to consider a transformation language to be defined by a metamodel, at level M2 of the metamodeling stack. This is also well accepted and many advantages are becoming apparent. In particular it is now clear that this choice allows to easily define and apply Higher-Order Transformations (HOTs), i.e. transformations taking other transformations as input and/or producing other transformations as output. AtlanMod has pioneered the use of HOTs in model engineering and this practice is now well accepted  .
The success of the model transformation paradigm has allowed many successful applications, and in turn generated a new set of constraints on the tools. For example the size of models and metamodels was previously limited when the main application was generating code from human-generated models like UML models. Now that we are moving towards model-driven reverse engineering and model-driven interoperability, we face huge or at least voluminous models that we have to process. One important activity of AtlanMod in 2009 has been to start investigating how the model-driven engineering do scale up in face of new constraints in term of size, of needed performance, of variability, etc. In the Lambda project, the team has lead a study on the identification of the dimensions of complexity in MDE. Additionally, AtlanMod contributed a case study  , as well as an accompanying solution  , on model transformation scalability to the GraBaTs'2009 workshop.