Background

Automated program repair (APR) aims to automatically repair software by modifying source code and testing changes against a test suite. The structure and representation of source code is an important ingredient in APR. To this end, we developed Program Repair Enhancement via Preprocessing (PREP), a pre-processing algorithm that modifies a program’s structure to make APR more successful.

Research Goals

This project will evaluate the benefits of PREP-based transformations for source code by taking datasets of programs and evaluating APR’s performance on unmodified programs vs. PREP-transformed programs.

Skills Needed

Familiarity with Linux, source code languages, and compilation.

Skills Gained

Linux, tool build, code evaluation.