Potential-fault cache-based regression test selection

Regression testing is the way to ensure the current version of the program is up and running. Continuous testing throughout the development cycle leads to detecting bugs as early as possible; however, it imposes huge overhead if the entire test suite has to be run. In this paper, powered by the goal of early detection of bugs, a regression test selection technique is proposed. The proposed technique uses a cache to keep the potentially error prone files and link them with test cases in a dynamic way that evolves with the evolution of code and test cases. Rather than running the entire test suite, only test cases that are linked with the most error prone files are selected and run on a regular frequent basis. Our cache outperforms previous related cache technique in terms of cache hit rate on 5 Apache projects. Based on the modified cache, the test selection module of the proposed technique can achieve a fault detection rate that exceeded 80% in all projects.