Run-time Code Generators for Model-Level Debugging in Domain-Specific Modeling

Code generators in the Domain-Specific Modeling (DSM) provide transformation from abstract models to specifica-tions that can be executed, interpreted, or compiled using a target language compiler. Modeling tools may include parsers assisting software engineers to perform an inverse task; i.e., to create a set of language concepts based on existing program code. Practical benefits from such a re-verse approach are limited, since source code cannot be used as an adequate base for creating language concepts on the high-abstraction level. Goal of application of the Model-Driven Development (MDD) approach, and DSM in particular, is to formally specify, using abstract models, a complete real system to the extent that models can be executed. In this paper we present an approach that completely achieve this goal. Model execution is implemented by the visual debugging of models, and submodels, which are dynamically created. In this approach, the most important role play performant code generators, so called run-time generators, and feedback that DSM tools get from run-time systems executing specifications.