A Comparative Study of Component Based Regression Testing Approaches without Source Code

Abstract: Context: Today, most of the software products are built with COTS components. When a new version of these components is available, it is difficult to perform testing as the vendors of the component do not usually provide source code. Various regression testing techniques have been developed, but most of the techniques are based on the source code for change identification. So, the testers are facing different challenges in performing effective testing. Objectives: The goal of this research is to find out the different approaches that are used to identify changes in modified COTS component, analyze the main characteristics of those approaches and investigate how these characteristics can be used in selection and development of CBRT approach. Methods: To fulfill the aims of the research, we have conducted systematic literature review of different CBRT approaches from the year 1993-2010. From systematic literature we found out 32 papers relevant to our study. Data related to our research are extracted from those papers and conclusion is made. The relevant articles were searched in six scientific databases such as IEEE Explore, ACM Digital Library, SpringerLink, Science Direct, Scopus, and Engineering Village. Furthermore, online survey was conducted based on the characteristics of CBRT approaches. This survey was conducted to validate the SLR result. Results: From the systematic Literature Review we have found out 8 different characteristics of CBRT approaches such as applicability, automation, complexity, behavior model used, coverage criteria, strength and weakness, theory used and input. We observe that these are the most important characteristics in CBRT approaches and these approaches should be considered in selecting or developing new CBRT approach. The results from the survey also validate our findings. From survey some more factors were identified. Conclusion: The research develops the state-of-art of CBRT approaches towards future research. The result of this thesis will be helpful for the researchers as well as practitioners who are working on CBRT. The result of the thesis can be considered as a basis for further study. Based on the result of this thesis further study can be done on making a framework based on these characteristics and support component based regression testing.