Finding Bugs in Program Generators by Dynamic Analysis of Syntactic Language Constraints1

Conference Paper (2014)
Author(s)

S.T. Erdweg (Technische Universität Darmstadt)

Vlad Vergu (TU Delft - Programming Languages)

Mira Mezini (Technische Universität Darmstadt)

Eelco Visser (TU Delft - Programming Languages)

Research Group
Programming Languages
DOI related publication
https://doi.org/10.1145/2584469.2584474
More Info
expand_more
Publication Year
2014
Language
English
Research Group
Programming Languages
Pages (from-to)
17-20
ISBN (print)
978-1-4503-2773-2

Abstract

Program generators and transformations are hard to implement correctly, because the implementation needs to generically describe how to construct programs, for example, using templates or rewrite rules. We apply dynamic analysis to program generators in order to support developers in finding bugs and identifying the source of the bug. Our analysis focuses on syntactic language constraints and checks that generated programs are syntactically well-formed. To retain a language's grammar as the unique specification of the language's syntax, we devised mechanisms to derive the analysis from the grammar. Moreover, we designed a run-time system to support the modular activation/deactivation of the analysis, so that generators do not require adaption. We have implemented the analysis for the Stratego term-rewriting language and applied it in case studies based on Spoofax and SugarJ.

No files available

Metadata only record. There are no files for this record.