The goal of this thesis is to study challenges in constructing call graphs in dynamic programming languages such as Python and JavaScript and to implement a framework to study the soundness of call graphs for such languages.

Call graphs are foundational static analyses and a prerequisite for any interprocedural analysis. Constructing sound call graphs is hard for real-world programs due to challenging features of programming languages and ecosystems.

The thesis should study the challenges that dynamic languages pose to call-graph construction. Based upon this study, a benchmark of test cases for individual challenges shall be created and a framework implemented that uses these benchmarks to study the soundness of existing call-graph implementations in various tools. This framework could be incorporated in our existing JCG tool, that performs similar tasks for Java Bytecode programs.

Examiner: Prof. Dr.-Ing. Mira Mezini