Hi Lofi, On 8/12/05, Lofi Dewanto <[EMAIL PROTECTED]> wrote: > it's a pitty that I haven't finished my article about AndroMDA and MTL > yet... But I'm quite sure that I'll finish it very soon (because all the > example codes are already finished...). I heard from Matthias that Chad is collaborating with the ATL team as well. I think each one of us should write out his ideas about an architecture that supports multiple model transformation languages. Olaf Muliawan will take the lead for MoTMoT, I'll collaborate with him and integrate my work on CAViT (which makes us of MoTMoT, see http://www.win.ua.ac.be/~pvgorp/research/#cavit) as well.
> I like the idea of the ability to "plug" (transformation chainning) > model-model transformation languages into AndroMDA (MTL, ATL, ModTransf, > ...). With this style we will be able to use the right tool for the > right job. Since each model-model transformation language has its own > advantages and disadvantages just like common programming languages > (imperative, declarative, function-oriented, graph-oriented, ...). Absolutely! We should integrate all our open source model transformation tools. AndroMDA seems to be an interesting platform for this effort thanks to its large user base and amount of mature cartridges (that can be refactored to AndroMDA 4 and then compared in flexibility). > The ultimate target is that we can reach "a model-driven transformation > language" (model-model and model-text) with AndroMDA (this is what > MotMot tries to reach if I'm not wrong... Peter, could you explain > this?). So, we can use UML to define our transformation and just use the > "correct" cartridges (ATL, MTL, ModTransf, etc.) to generate all or some > of the transformation codes needed. When MoTMoT is integrated in AndroMDA, the result will indeed allow a smooth transition from model-to-model to model-to-text. > So, we can use UML to define our transformation and just use the > "correct" cartridges (ATL, MTL, ModTransf, etc.) to generate all or some > of the transformation codes needed. In MoTMoT, indeed you use UML to define your model (to model) transformations. More precisely, MoTMoT is based on a UML profile for Story Driven Modeling (SDM), a language based on controlled graph rewriting. Don't feel overwhelmed by the theoretical connotation of "graph rewriting" ;-) It's really powerful but very easy to learn. Your suggestion to use ATL, MTL, ModTransf or others to translate the UML describing the model transformations to transformation code is reasonable. In fact, we've had a master thesis about translating the UML into ATL by means of ATL transformations. However, the student failed as the transformation is not so straightforward as the examples bundled with ATL. Therefore, MoTMoT still uses AndroMDA 3.0's direct model-to-text approach (given the MMFs). If you're interested in applying MTL for MoTMoT's complex SDM to JMI transformation, please let us know. We're still looking for better solutions, including the application of MoTMoT on itself etc. Olaf Muliawan is currenly working on the MoTMoT documentation but we still have to review it. In the meanwhile, please refer to my CAViT paper to see how SDM can be applied for vertical transformations: http://www.win.ua.ac.be/~pvgorp/research/#cavit CU, -- Pieter. ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ Andromda-devel mailing list Andromda-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/andromda-devel
