Language Engineering for Analyzable Executable DSMLs

The project LEA-xDSML (Language Engineering for Analyzable Executable Domain- Specific Modeling Languages) resides in the context of Model-Driven Engineering (MDE), which proposes the use of domain-specific modeling languages (DSMLs) to reduce the com- plexity associated with the development of complex software-intensive systems, as, for in- stance, found in the automation domain, production domain, and automotive domain. DSMLs are increasingly being developed to continuously leverage the domain-specific ex- pertise of the various stakeholders involved in the development of complex system. Thereby, the integration of domain-specific knowledge into DSMLs can significantly improve the productivity of the development process and the quality of the final system. However, the development of DSMLs has also been recognized as a challenging and significant software engineering task itself. In this project, we focus on the challenges associated with the development of executable DSMLs (xDSMLs) that support the modeling and analysis of complex system behaviors through model execution. In particular, we aim at overcoming the following three challenges: the lack of foundations for formalizing xDSMLs in a way that allows for model-level anal- yses; the high efforts associated with the development of domain-specific analysis tools for xDSMLs; and the lack of fault localization techniques for efficiently identifying faults in models defined with xDSMLs. To overcome these challenges, the aim of this project is to develop a novel engineering framework for xDSMLs that will provide (i) concepts, techniques and processes to formalize xDSMLs usable for model-level behavior analyses, (ii) automation techniques for efficiently developing domain-specific model analysis tools for xDSMLs, (iii) and fault localization mechanisms for xDSMLs that allow an efficient debugging of models. The framework will be iteratively developed and evaluated. The methodology for evaluat- ing the framework builds on three major pillars, namely case studies, experiments with our master students (around 100 students every year), and collaborative studies with international collaborators. The results of the project will significantly ease the development of xDSMLs and accom- panying model analysis tools, and thus lead to reduced development costs of xDSMLs and at the same time increased quality of systems developed with xDSMLs. This will present a ma- jor cornerstone in the model-based development of complex software-intensive systems.

Developing software-intensive systems is still a major challenge as current systems have to incorporate complex domain knowledge in order to realize true benefits. Model-Driven Engineering (MDE) aims at reducing the accidental complexity associated with the development of complex software-intensive systems through the use of domain-specific modelling languages. Such languages are designed to allow the development of systems by domain experts. However, to realize their full potential by incorporating domain experts also for later phases in the software development lifecycle, automated analysis, testing, and debugging support are a must for such languages. Within the LEAxDSML project, we target the automated generation of analysis, testing, and debugging support for domain-specific modelling languages. Thus, the first research area of this project is to provide a systematic language engineering method comprising concepts, techniques, and processes that support language developers to identify, design, and implement analyzability concerns of domain-specific modelling languages. In the second research area, we aim at developing generic approaches that provide out-of-the-box testing and debugging tools for domain-specific modelling languages. Finally, the third research area of this project is to provide fault localization mechanisms that allow domain engineers to trace back incorrect behaviour to precisely those software elements that cause the fault. Having this extended tool support allows to engineer domain-specific modelling languages with less effort but having sophisticated environments to utilize such languages in many different domains.