All software evolves, and programming languages and programming language tools are no exception. And just like in ordinary software construction, modular implementations can help ease the process of changing a language implementation and its dependent tools. However, the syntactic and semantic dependencies between language features make this a challenging problem. In this paper we detail how programming languages can be implemented in a modular fashion using the Rascal meta-programming language. Rascal supports extensible definition of concrete syntax, abstract syntax and operations on concrete and abstract syntax trees like matching, traversal and transformation. As a result, new language features can be added without having to change existing code. As a case study, we detail our solution of the LDTA'11 Tool Challenge: a modular implementation of Oberon-0, a relatively simple imperative programming language. The approach we sketch can be applied equally well to the implementation of domain-specific languages. We present an extended language engineering case-study in the meta programming language Rascal.The case-study is the implementation of Oberon-0 (from the LDTA'11 tool challenge).The case-study illustrates the OO-like modularity features of Rascal, to support extensible syntax definition and extensible language operations.
Additional Metadata
Keywords language workbench, Rascal, language engineering, modularity
THEME Software (theme 1)
Publisher North-Holland
Stakeholder Unspecified
Persistent URL dx.doi.org/10.1016/j.scico.2015.11.003
Journal Science of Computer Programming
Project Domain Specific Languages: A Big Future for Small Programs
Citation
Basten, H.J.S, van den Bos, J, Hills, M.A, Klint, P, Lankamp, A.W, Lisser, B, … Vinju, J.J. (2015). Modular Language Implementation in Rascal -- Experience Report. Science of Computer Programming, 114, 7–19. doi:10.1016/j.scico.2015.11.003