RT-Selection : A Regression Test Selection Technique Using Textual Differencing and Change Impact Analysis

Regression testing intended to provide confidence that newly introduced changes do not obstruct the behaviours of the existing and unchanged parts of the software. One of simple and basic regression testing techniques is retest-all, but it requires lot of time and cost. The regression test selection technique selects a subset of previous test cases to retest the changed software. It, therefore, can reduce the time and cost through reducing the number of test cases for regression testing. This paper proposes a new regression test selection technique, RT-Selection, which can perform the regression test more effectively than the existing selection techniques. It can be explained in two approaches. First, it uses textual differencing to fine change. Second, it uses change impact analysis to fine the software riffle to trace test cases. RT-Selection helps testers efficiently select a subset from previous test cases for regression testing. We also propose 4 Guidelines and inference rules to support this technique. Guidelines and inference rules help testers can perform this technique more systemically. We performed a case study to show feasibility of the proposed technique with graduate and undergraduate software engineering classes in Konkuk University.