Automated Repair of Multi-fault Programs: Obstacles, Approaches, and Prospects

Modern automated program repair (APR) tools are well-tuned at repairing single fault programs (i.e., programs in which only one fault can occur at time). However, real-world software projects typically contain multiple bugs at the same time, which can interact with and mask each other in a variety of ways. The complex interaction of faults in multi-fault programs makes the automated repair problem more challenging than the traditional practice of presuming that a program contains a single fault. This paper studies the repair problem of multi-fault programs and identifies the main obstacles that arise when handling such programs using current repair approaches. The paper also describes three repair approaches for multi-fault programs, namely iterative, parallel, and simultaneous. While the simultaneous repair strategy depends on using cutting-edge fault localization techniques that enable the APR approaches to locate many faults at once, the iterative and parallel repair approaches rely on adapting the existing repair techniques for single-fault programs to handle multi-fault programs. Finally, the paper discusses each approach's advantages and drawbacks as well as the conditions in which the approach can be used successfully. To our knowledge, this is the first paper to specifically study and address the repair problem of multi-fault programs.