Regression test selection for JAVA using static analysis

Abstract

Modern software development demands continuous maintenance of software so that it does not become obsolete. In fact, developers frequently modify, add or delete code. When such code changes occur, it is best practice to regression test the program. This is a simple process involving the execution of the project's test suite in order to identify system faults which might have resulted due to code modifications. The problem with this approach is that, if the test suite takes a long time to run, the developer wastes a significant amount of time rerunning the test suite, which could otherwise be invested in doing something more productive in the project. Consequently, a regression test selection technique may be used in order to reduce this execution time by selecting a subset of tests from the project's test suite in order to test modified code. This Final Year Project designed, implemented and evaluated a regression test selection technique for Java programs through the use of static analysis of code. This technique is capable of choosing a subset of unit tests with the attempt of decreasing the test execution time while testing modified code portions as well as code parts affected by such modifications. This technique has a high measure for recall and precision because it is capable of choosing the majority of tests which exercise the modified code. This behaviour may result in considerable reductions in a program's test suite size. Moreover, our regression test selection technique is quite efficient if applied to adequate programs. Although techniques with a higher measure for recall and precision have been developed, through this study it has been concluded that the use of selective regression testing could result in significant benefits to a software development team. However, such benefits are dependent on several software factors like software size, the amount of modifications, test suite's execution time and coverage obtained by the tests.