Research Of A Test Suite Reduction Method Based On Process Slicing

Software testing throughout the entire process of software development, it's the longest and most expensive stage in software development process. Software testing carries on by running test cases with the purpose to detect the potential errors in software program. Then provide some necessary basis for the developers or maintenance personnel. Software testing must discovery more mistakes, the testing process which can find errors is valuable.With the escalation of the software version and the expanding of software, software testing process required for more and more test cases, to operation and maintenance these test cases need some cost, the expand of test cases will bring the increasing of cost to operation and maintenance them. However, in the these test cases, there may be a great number of redundant test cases, the presence of them in the software testing is meaningless, we can delete these redundant test cases to reduce software development costs, that is, we can find a subset of the test cases to finish the testing of source code, this test suite must be have the same test coverage with original test suite. There are many methods used for reduction of test cases, including many heuristic algorithms:such as greedy algorithms, HGS algorithm, GRE algorithm, GE algorithm, experiments prove that these methods can reduce the test cases effectively. With the reduction of test cases, followed by a question that is with the redundant of test cases, the error detection capability of reduced test cases will also be reduced. How to ensure the error detection capability of test cases based on the reduction of the original test suite is the main content of this article.After analyze and summarize the archived results, we bring in the thinking of program slice, to find out the sentences as key points which may cause program faults but test cases may not detect and the results of testing is not good. From the key points to slicing the program by constructing the slicing dependence graph which discussed in this paper, in this way can find out the range which will cause the key points'errors. Then use this range as the test criteria in choose and remaining redundant test cases, in order to keep the fault detection ability of the reduction test cases maintain a certain level. Finally, by doing experiments to prove this method is effective in maintaining fault detection ability and available of test cases. That is achieving the purpose to reduce the test suits and maintaining the fault detection capability of reduced test cases.In this paper will complete the following aspects of work:1. Define key points for operational procedures and given the classification for the operation procedures.2. Presents a new construct method of dependence graph for program slicing.3. Analyze the input error range of the key points, and use this range as a test suite reduction coverage criterion.4. Turn the thought of choose and remaining redundant test cases to algorithm, and improve it with the proposed coverage criteria.5. Design two simulation experiments combined with the contents of this paper, the first is used to insure the dependency graph does not affect the correctness of program slicing, and another is used to validate the method which researched is effective in maintaining fault detection capability of reduced test suits. It can achieve the purpose to reduce the test suits and maintaining the fault detection capability of reduced test cases.