Efficient Regression Test Selection and Recommendation for Component Based Software

Abstract

Component-Based Development is an approach of developing software systems by using components. Component-based software system may contain external components as well as in-house built components. During the maintenance phase the components get altered or modified very often .When a component is altered, it is not only that component which is affected but it is the whole system which is affected. The type of testing which not only ensures that the modified component is working fine but also ensures that the changes have no adverse or severe effects on the rest of the system is called as regression testing. But due to unavailability of the knowledge about third party components it is difficult for the component users or testers to perform the testing in an efficient manner. As the Component users don’t have the information regarding the modifications done in the component, it creates a problem for them to do an appropriate selection of the test cases from the original test suite for testing the altered system. Thus there is a need of an efficient regression test selection approach which results in a reduced regression test suite. An efficient regression test selection and recommendation for component based software approach is proposed. The approach – “Regression Test Selection and Recommendation” (RTSR) uses UML diagrams (state chart diagrams and sequence diagrams) to investigate the changes, which are further used to select a subset of test cases from the initial test suite and also used to recommend new test cases for regression testing of component based software. The approach RTSR identifies the change, selects test cases from the original test suite, also recommends new test cases if needed and gives regression test suite as final output. An application is designed to validate the approach using case study of automated teller machine. The experimental results demonstrate that RTSR selects lower number of test cases than original test suite and a percentage reduction of 61.9 % is achieved in the regression test suite and also all the changes are tested successfully proving that the approach is efficient.