Lightweight Fault Localization Combining with Fault-Context

Abstract

Lightweight fault localization technique is a popular class of automated approach to assist programmers in debugging, which often outputs an ordered list of program entities sorted based on their likelihood to be the root cause of a set of failures. However, the technique only focus on calculating the association between program entity and failures, and did not consider program entity's fault context may influence the result. A lightweight fault localization based on fault-context was proposed, which combine suspiciousness of program entity and suspiciousness of program entity's fault-context. We conducted an experiment in which our approach was applied to seven benchmark programs. The experimental results show that our approach combining DStar and fault-context can improve absolute ranking with effective rate of 34.8% for 132 faulty versions from the seven benchmark programs. It is worth mentioning that our approach can obtain average improvement of 65.18% for those improved program if those is effective to SFL, and root causes of failures of 7 buggy programs were improved ranking at the top in the fault ranking report.