Towards a general foundation for formalism-specific instrumentation languages

Experimenters need to configure the data collection performed during a simulation run, as this avoids overly large output data sets and the overhead of collecting them. Instrumentation languages address this problem by allowing experimenters to specify the data of interest. Such languages typically focus on a specific modeling formalism. While this allows for a more expressive syntax, it also prohibits their application to other formalisms. To resolve this trade-off, we propose a formalism-independent model of the instrumentation semantics and use it as a basis for developing embedded domain-specific languages (DSLs). Our instrumentation DSLs share common code, allow to add formalism-specific syntax, and are easy to extend.