Patterns and styles for incremental model transformations

Abstract

The usability of code generators in industrial modeling frameworks is frequently hindered by long re-generation time imposed by complex dependencies between di erent modules and chains of code generation phases. Incremental model transformation techniques may improve both scalability and usability by regenerating only speci c design artifacts dependent on a speci c change. In this paper, I provide an overview on di erent styles and levels of incrementality in model transformations of the Viatra open source framework.

Keywords

• model transformation patterns
• incremental evaluation
• reactive programming
• code generation

-

each model change [ 10,9,2 ]. Furthermore, in incremental code generators, only artifacts a ected by a speci c model change will be re-generated while the rest of the target output is left unaltered. Such techniques have been successfully used in various design and validation tools for embedded and cyber-physical systems as reported in [ 4,8,5 ]3. 2

Di erent styles of incrementality The actual level of incrementality in these applications di ers and it can be netuned for a transformation problem. In our practice, di erent styles and levels of incrementality could be categorized as follows: { No incrementality is a batch transformation re-executed from scratch for all source models and all changes of these source models. { Dirty incrementality is a large-step incrementality approach which is very common in industrial practice. Here one marks a model to be dirty upon a source change, and then re-run transformations only on dirty models. This technique signi cantly reduces the number of regenerated artifacts, but cleanup after an error and chaining of transformation steps is non- trivial. { Incrementality by traceability is a small-step incrementality approach which relies on the existence of traceability links between source and target model elements generated during a rst transformation. Missing trace links are detected and incrementality is achieved by re-executing the transformations for untraceable elements only. This technique may further reduce the e ects of a change, but it depends on a smart matcher of traceability links. { Reactive source incrementality is also a small-step incrementality approach where the ring of transformation rules are triggered only by changes in the source model but not the traceability links between source and target models. Such approaches may restrict the expressiveness of the transformation language, but o er improved chaining of incremental transformations.

Sample model transformations illustrating these scenarios are available from https://github.com/IncQueryLabs/incquery-examples-cps with complete source code, documentation and performance evaluation using the Viatra framework. More details on reactive incremental transformations are provided in [ 11 ]. Further challenges. In practice, we nd the complex interactions and chaining of incremental transformation and code generation steps to be the most challenging. In this setup, a change in a (front-end) source model may trigger reactions, which results in changing one or more target models. However, as these model changes may trigger further reactions along the transformation chain which require incremental handling themselves - potentially using a di erent style. With inappropriate tool support, it is surprisingly easy to achieve circular dependencies between di erent steps where di erent incremental transformations are executed continuously (e.g. simultaneously running validations and quick xes). 3 And also in https://github.com/IncQueryLabs/EMDW-Common/wiki Acknowledgments. The author is highly grateful to the entire Viatra development team, and particularly, Gabor Bergmann, Abel Hegedus, Akos Horvath, Istvan Rath and Zoltan Ujhelyi for di erent sort of inputs and feedback. This paper was partially supported by the MONDO Project (EU ICT-611125), the FP7 ARTEMIS CONCERTO (ART-2012-333053) project and the MTA-BME Lendulet Research Group on Cyber-Physical Systems.

References

1. Bergmann , G. , David , I. , Hegedus, A. , Horvath , A. , Rath , I. , Ujhelyi , Z. , Varro , D. : Viatra 3: A reactive model transformation platform . In: Theory and Practice of Model Transformations - 8th International Conference , ICMT 2015 . LNCS , vol. 9152 , pp. 101 { 110 ( 2015 )

2. Cabot , J. , Teniente , E.: Incremental integrity checking of UML/OCL conceptual schemas . J. Syst. Softw . 82 ( 9 ), 1459 { 1478 ( 2009 )

3. Etzlstorfer , J. , Kusel , A. , Kapsammer , E. , Langer , P. , Retschitzegger , W. , Schoenboeck , J. , Schwinger , W. , Wimmer , M.: A survey on incremental model transformation approaches . In: Proceedings of the Workshop on Models and Evolution co-located with MoDELS 2013 . CEUR Workshop Proc., vol. 1090 , pp. 4 { 13 ( 2013 )

4. Horvath , A. , Hegedus, A. , Bur , M. , Varro , D. , Starr , R.R. , Mirachi , S. : Hardwaresoftware allocation speci cation of ima systems for early simulation . In: Digital Avionics Systems Conference (DASC) . IEEE, Colorado Springs , US ( 2014 )

5. Horvath , A. , Rath , I. , Hegedus, A. , Balogh , A. : Decreasing your co ee consumption with incremental code regeneration . In: EclipseCon France ( 2015 )

6. Jouault , F. , Tisi , M. : Towards incremental execution of ATL transformations . In: Tratt, L. , Gogolla , M. (eds.) Theory and Practice of Model Transformations, LNCS , vol. 6142 , pp. 123 { 137 . Springer Berlin Heidelberg ( 2010 )

7. Kolovos , D.S. , Rose , L.M. , Matragkas , N. , Paige , R.F. , Guerra , E. , Cuadrado , J.S. , De Lara , J. , Rath , I. , Varro , D. , Tisi , M. , Cabot , J.: A research roadmap towards achieving scalability in model driven engineering . In: Proceedings of the Workshop on Scalability in Model Driven Engineering . pp. 2 : 1 {2: 10 . BigMDE '13, ACM ( 2013 )

8. Micskei , Z. , Konnerth , R. , Horvath , B. , Semerath , O. , Voros, A ., Varro , D. : On open source tools for behavioral modeling and analysis with fuml and alf . In: Proceedings of the 1st Workshop on Open Source Software for Model Driven Engineering (OSS4MDE@MoDELS 2014 ). CEUR Workshop Proceedings , vol. 1290 , pp. 31 { 41 ( 2014 )

9. Reder , A. , Egyed , A. : Incremental consistency checking for complex design rules and larger model changes . In: Model Driven Engineering Languages and Systems, LNCS , vol. 7590 , pp. 202 { 218 . Springer Berlin Heidelberg ( 2012 )

10. Ujhelyi , Z. , Bergmann , G. , Hegedus, A ., Horvath , A. , Izso , B. , Rath , I. , Szatmari , Z. , Varro , D. : EMF-IncQuery: An integrated development environment for live model queries . Sci. Comput . Program. 98 , 80 { 99 ( 2015 )

11. Varro , D. , Bergmann , G. , Hegedus, A ., Horvath , A. , Rath , I. , Ujhelyi , Z. : Road to a reactive and incremental model transformation platform . Software and Systems Modeling ( 2016 ), in press