History-Based Test Case Prioritization for Failure Information

Abstract

From regression tests, developers seek to determine not only the existence of faults, but also failure information such as what test cases failed. Failure information can assist in identifying suspicious modules or functions in order to fix the detected faults. In continuous integration environments, this can also help managers of the source code repository address unexpected situations caused by regression faults. We introduce an approach, referred to as AFSAC, which is a test case prioritization technique based on history data, that can be used to effectively obtain failure information. Our approach is composed of two stages. First, we statistically analyze the failure history for each test case to order the test cases. Next, we reorder the test cases utilizing the correlation data of test cases acquired by previous test results. We performed an empirical study on two open-source Apache software projects (i.e., Tomcat and Camel) to evaluate our approach. The results of the empirical study show that our approach provides failure information to testers and developers more effectively than other prioritization techniques, and each prioritizing method of our approach improves the ability to obtain failure information.