Design of Test Suits Using Program Slicing

Abstract

A major portion of the software development costing goes to testing of the software and in particular regression testing. This is because regression testing is generally done on modified program by re-execution iof iselected itest icases iwhich indicate that ithere iare ino new faults in the software. The major problem in most testing techniques is size of the source code. Therefore, there is need to have a testing method which can effectively minimize the inumber iof ithe itest icases ithus ireducing both icost iand itime and also maintain the effectiveness of the software. To address above said problems, Regression iTest iSelection i(RTS) i, Test iSuite iMinimization i(TSM) iand iprogram slicing have been used ito ireduce ithe iamount iof itesting ithat ineeds ito ibe iperformed. iTesting effort is reduced by re-running only some of the available tests. A igreat ideal iof iresearch ieffort ihas ibeen spent on finding an efficient approach to regression testing. Researchers have investigated various approaches for addressing regression testing problem. Some of them are: RTS, TSM and Hybrid approach etc. Firstly, we propose a multi-criteria Fuzzy rule based method for determining probability of selection of test cases. iThe second technique Row-Column Reduction technique eliminates redundant test cases in newlineminimum iexecution itime ifrom ia itest isuite iby covering iall ithe ifaults. Third, a new variable based technique called Hybrid Technique for selecting and minimizing test cases has also been proposed. . In this thesis, one technique has been proposed for selection of test cases namely a multi criteria fuzzy rule based method. Two techniques Dom Red and Row-Column have been proposed for minimization of test cases and a hybrid technique using slicing and firefly has been proposed for both selection and minimization of test cases. All the above proposed techniques have been validated by applying on various programs and the results obtained were found to be satisfactory in comparison to the existing approaches. newline