Language Composition via Kind-Typed Symbol Tables

By: Arvid Butting, Judith Michael, Bernhard Rumpe

Abstract

The modularization of domain-specific modeling languages (DSMLs) fosters individual reuse of DSMLs in differentcontexts. Within this article, we discuss how it is possible to refer to model elements of other languages when composing different DSMLs. Related approaches usually rely on a DSML-agnostic language infrastructure that tests the compatibility of model elements via types encoded in Strings without any consistency checks. We propose the "strongly kind-typed" symbol table as an extension to the compiler approach to integrate the syntax of the languages using symbol tables that assign a symbol kind to each name definition. Our approach can be integrated into language workbenches that provide a symbol table infrastructure as part of a DSML implementation. The kind-typed symbol tables are integrated into the language workbench MontiCore. Strongly kind-typed symbol tables utilize the type system of the language workbench’s host language to ensure type consistency between the language-specific symbol table infrastructures during DSML composition, which ultimately supports DSML engineering in the large.

Keywords

Software Language Engineering, Symbol Tables, Language Composition.

Cite as:

Arvid Butting, Judith Michael, Bernhard Rumpe, “Language Composition via Kind-Typed Symbol Tables”, Journal of Object Technology, Volume 21, no. 4 (October 2022), pp. 4:1-13, doi:10.5381/jot.2022.21.4.a5.