In several projects, we have been building automotive toolchains based on EMF models and model transformations with Xtend. In this talk, we will introduce our "lessons learned" from projects, in which we integrate automotive engineering data from different sources (relational databases, specific configuration files, EMF) into a consolidated model for product lines of electronic control units (ECUs) for cars and then transforming it to AUTOSAR. These models easily exceed a size of 1mio elements.
- Xtend's special support for model-to-model transformation (i.e. create methods)
- Specific extensions with Xtend's active annotations (caching, logging)
- Tracing the transformations to the end model - each final model in the output retains information about the original sources and the transformations applied to it.
- Generation of specific HTML documentation for the transformation, making it easily visible which model elements are produced / consumed by which transformations
- Traceability: Using Xtend's active annotation to keep traceability information from the implemented code to the specification in Word documentation.
- Performance analysis
- Short comparison to other model-to-model transformation technologies (e.g. QVTO)
After attending the talk, the participants will have a bag of new ideas and methods to improve the design and implementation of model-to-model transformations in their projects.