Fault Localization Guided Execution Comparison for Failure Comprehension

Fault Localization has been the research focus in past decades. However, recent studies have shown that automated fault localization is not fit for realistic software debugging progresses, thus an attractive idea is how to employ the results of fault localization techniques to assist software debugging. During software debugging processes, programmers always need to re-execute several test cases after confirming the fault and then diagnose the fault by means of analyzing and comparing the context. Therefore, the crux of fault diagnosis is how to select test cases. In other words, how to choose test cases that could reveal the fault from a big test suite, which is more helpful for the programmers to understand and fix the fault more quickly and effectively? In this paper, we intend to integrate the result of fault localization to characterize the ability of revealing fault of test cases and provide two test case selection strategies, which will assist the programmers with understanding the fault more pointedly.