Change impact analysis and test case management for regression testing of object-oriented programs

Abstract

During software maintenance phase, programs are evolved for various reasons; to adapt programs to new operational environments, to correct their errors, to enhance their functionalities, or to ease future maintenance. Whenever a program is modified, it must be retested to ascertain whether changes have been made correctly and whether those changes have caused any adverse effect on its behavior. Regression testing is the activity of validating that the modified program is still right. Since testing is a complicated and expensive activity, most regression testing techniques attempt to identify only those parts of the program that are affected by a change and retest them. This activity is called change impact analysis. Moreover, they try to reuse test data obtained from the previous testing phase in order to reduce the time and effort of retesting. This dissertation describes an environment of regression testing for C++ programs. Especially, it focuses on change impact analysis and test data management. Impact analysis technique identifies which parts should be retested after a software system is modified. We are interested in identifying the impacts of changes at the class member-level by using dependency relations among class members. We try to find out which member functions need unit-level retesting and which interactions between member functions need integration-level retesting. To get precise analysis results, we adopt a technique that classifies types of changes and analyze the impact for each type. Primitive changes, changes which are associated with C++ features, are first defined and their ripple effects are computed in order to construct a firewall for each type of changes systematically. Test data history is maintained at the class member-level and test data that execute program components in the computed firewall are selected as reusable ones. We have applied our prototype tool to a real system with small size. This case study shows some evidence that...