Fix Cache Based Regression Test Selection

Abstract

Regression testing is a crucial step in the software development process, which ensures the quality of software systems by detecting whether new faults have been introduced into previously tested code. Regression testing becomes costly as more and more regression test cases are created. Regression test selection, which selects a sub-set of the available regression test cases based on different criteria, is a well-known method to reduce test scope and improve the efficiency of regression testing. This paper presents a new method, named fix cache based regression test selection, which computes test case coverage based on what files were updated to fix faults found by the test cases. Our method uses a cache to monitor the most fault-prone files and recommends test cases related with continuously updated files. The method is useful for predicting new faults and selecting the most fault-prone test cases for automatic regression testing. The thesis explores the concepts and processes for how to implement and evaluate this method. We have implemented the method and evaluated it during two months‘ period in a large, industrial, embedded, real-time software system. Our results show that the fix cache based selection method is effective with reaching weekly cache hit rates in the range 50%-80% for a fully automatic regression testing.