Based On The Test Case Priority Automated Regression Testing Research

Regression Testing plays an important role in software testing. Automated regression testing can efficiently simplify the tedious process of regression testing. However, in some huge and complex software projects, regression testing can’t finish all time-consuming test cases in limited time. So it will be very low performance.Aim at how to improve automated regression testing’s efficiency, the author had firstly studied three approaches including test cases selection, minimum and prioritization, and conformed to adopt test cases prioritization method. Then the author had considered and balanced several prioritization technologies (no prioritization, random prioritization, optimal prioritization, code coverage and FEP prioritization). Finally the author had come up with test cases covering impacted functions prioritization technology (code coverage prioritization).In this thesis, Test cases prioritization involves1) source code function instrumentation:instrument into source code to get information in GCC compiling phase;2) database design, collecting test cases and function information into database;3) test cases clustering, clustering similar, relevant test cases;4) patch parsing, parsing patch to get impacted functions list;5) test cases priority selection, selecting test cases according to impacted functions.Finally, the author had implemented all of above modules correctly and applied test cases prioritization into Intel Linux Graphics’automated regression testing project.This essay gives the test results for both original and new test cases prioritization system, we can get below conclusions.1) When functions traced by less test cases are changed, new system can successfully select regression test cases with large probability (over90%);2) When functions traced by most test cases are changed, new system can select parts of regression test cases (but we can adapt some auxiliary strategies to improve this poor situation). However, from the daily testing, this new system can select regression test cases with84%probability. So we can get this new system can indeed improve automated regression testing efficiency.