Prioritization of Re-executable Test Cases of Activity Diagram in Regression Testing Using Model Based Environment

Abstract

As we all know, software testing is of vital importance in software development life cycle (SDLC) to validate the new versions of the software and detection of faults. Regression Testing, however concentrates on generating test cases on changed part of the software to detect faults more earlier than any other testing practices. In case of model based testing approach, testing is performed using top-down method (black box method) and design models of the software, for example, UML diagrams. UML diagrams gives us requirement level representation of the software in graphical format which is now a days a standard used in software engineering. In our proposed approach, we have derived a new technique which has never been used before to prioritize the test cases in model based environment. In this technique, we have used activity diagram as an input to the system. Activity diagram is used basically because it gives us the complete flow of each and every activity involved in the system and represents its complete working. Activity diagram is further changed as the requirement changes, each time, when the changes happen, they are recorded and test cases are generated for the changed diagram, test cases are also generated for the original diagram. Test cases for both the diagrams are compared and classified as re-usable and re-executable test cases. Re-usable test cases are those that remain unchanged during requirement changes and re-executable test cases belong to the changed part of the diagram. Then re-executable test cases are prioritized using one heuristic algorithm based on ACT(Activity Connector) table. Now, the question is why to prioritize only the re-executable test cases. Because, any how we have to execute re-usable test cases, as they remain same for both the versions of the diagram and are already tested when original diagram was made. But, re-executable test cases are never been tested and may detect faults in the modified design quickly and by prioritizing them we can also reduce the execution time of the test cases which will give us effective testing performance and will evolve a better new version of the software. All the existing prioritization techniques are either code based or are using various tool supports. Code based techniques are too complex and tedious because for a small change in code, we need to test whole application repeatedly. And in case of tool support, we have multiple assumptions and constraints to be followed. This proposed technique will surely give better results and as the type of technique has never been used before will also prove very effective.
DOI: 10.17762/ijritcc2321-8169.150771