Formalizing DSL semantics for reasoning and conformance testing

A Domain Specific Language (DSL) focuses on the essential concepts in a certain problem domain, thus abstracting from low-level implementation details. In combination with code generators, DSLs bring software development closer to domain requirements. The development of DSLs usually centers around the grammar and a code generator; there is little attention for the semantics of the DSL. However, a formal semantics is essential for reasoning about specifications in terms of the DSL (i.e., DSL instances). We argue that the semantics should be expressed independent of a code generator. Thus semantic issues can be revealed that could otherwise remain undetected. We also use the semantics to define the conformance of an implementation to a DSL instance, and to automatically test conformance of the (generated) implementation code to a DSL instance. We illustrate our approach using an industrial prototype DSL for collision prevention. Keywords: Domain Specific Language (DSL); Semantics; Code Generation; Model-based Testing; Conformance Testin