navigation |
Date: 9 and 10 of March, 2015Presenters:
Organiser:
Objectives and Format
The objective of the workshop was to bring together the NECSIS consortium and fortiss (Germany), represented by Bernhard Schaetz. The goal was to identify common areas of research and potential synergies. Ericsson (Canada) was also represented at the workshop by Francis Bordeleau.
The format of the workshop was relatively loose, including presentations of around 1h during which discussions happened whenever they helped the audience have a better understanding the subjects being presented. During the second day discussions were around in-depth technical topics as well as topics regarding the present and future of NECSIS.
Summary of the Discussions and Future WorkTechnical discussionsThree technical topics were identified that are of main interest to most of the researchers that participated in the workshop: product lines, design space exploration and model transformations. Given the available time, only product lines and design space exploration were discussed.Product Lines: Levi started the discussion by introducing the product-line lifting work and the most recent application of this work to lifting a DSLTrans transformation. Bernhard then mentioned automotive companies in Europe use pure::variants and connect it to Doors and other tools - however this does not work well with Matlab/Simulink models. Bernhard also mentioned working with 150% models (models that contain elements from all variants) means constantly switching between two worlds. A separate product-line manager is responsible for updating the product line to accommodate changes, which means there is no first-class support for working with product lines directly by the developers. Also, automotive companies in Europe do not use inter-feature constraints to validate configurations. Francis mentioned some projects use model marking to handle product lines - meaning different overlap profiles that annotate the model for different variants. Michal suggested exploring the idea of a virtual platform, where "platform" is the set of reusable assets that can be configured into a product. The idea is that it is possible to use methods for identifying features and managing them in a lightweight way initially, and then gradually migrate toward more mature product line management approaches. Bernhard explained that at Fortiss ideas are being explored along these lines. According to Bernhard, this may be a good context for future collaboration. Design Space Exploration: Bran suggested using Marte, meaning the relevant subset of the profile must be found. Bernhard mentioned he is interested in building sub-profiles for Marte for particular design space exploration applications. Present and Future of NECSISOne of the major challenges regarding the NECSIS project is the integration of the software being developed by all the partners. This is not a simple task given that all groups develop prototypes on different platforms and with different technologies. Bernhard and Francis suggested for a potential new edition of NECSIS using a common platform from the start that would allow a lingua franca between the several groups to ease technological integration. Both also mentioned their experience with such platforms (AutoFOCUS 3 and Papyrus) that can be used to showcase prototypes that have been developed during an academic/industrial collaboration.Francis insisted on the benefits of open source for software develpment, mentioned the successful case of Papyrus and instigated NECSIS’s researchers to adopt it in their future work. Francis also mentioned that Ericsson has a strong interest in leveraging on open-source and that it is of interest for Ericsson if NECSIS plays in the open source arena. Bernhard expressed a strong interest in collaborating with a second edition of NECSIS, for example by applying for European funds for visits by European researchers to Canada or vice-versa. Another challenge faced by NECSIS researchers has been the access to models of software from GM. During his internship at GM, Bentley Oakes has begun developing the theory and software to automatically sanitize EMF models. Francis expressed interest in using this technology at Ericsson to ease the transfer of their own models to their academic partners. Both Francis and Bernhard suggested having a public report at the end of NECSIS, condensing the lessons learned during the project. Future WorkAs concrete future work following the workshop and the stay of Bernhard Schaetz in Montreal it has been decided that Bernhard, Levi and Bentley will work on a denotational semantics for the model transformation verification tool developed by Levi and Bentley. The plan is that such denotational semantics will ease the explanation of the current results of validity and completeness of the theory (currently formalized using operational semantics). |
Maintained by Levi Lúcio. | Last Modified: 2015/03/18 16:06:29. |