Evolution specification evaluation in industrial MDSE ecosystems

Domain-specific languages (DSLs) allow users to model systems using concepts from a specific domain. Evolution of DSLs triggers co-evolution of models developed in these languages. When the number of models that needs to co-evolve increases, so does the required effort to do so. This is called the co-evolution problem. We have investigated the extent of the co-evolution problem at ASML [1], provider of lithography equipment for the semiconductor industry. Here we have described the structure and evolution of a large-scale ecosystem of DSLs. We have observed that due to the large number of artifacts that require coevolutionary activity, manual solutions have become unfeasible, and an automated approach is required. A popular approach for automating co-evolution is the operator-based approach. In this paper we have evaluated the operator-based approach on a large-scale industrial case-study of twenty-two DSLs and 95 model-to-model transformations with a revision history of over three years, and have revealed deficiencies in existing operator libraries. To address these deficiencies we have presented a topdown methodology to derive a complete set of operators.