Towards Generating Local High-Coverage Tests for Evolving Software

Test suite augmentation techniques focus on identifying test requirements or creating new tests to adequately exercise the effects of program changes. To reveal the effects of changes as much as possible, this paper propose a test suite augmentation approach, a technique that aims at covering as many as possible longer dependence chains related to the changes in a cost-effective way. In particular, the scope of thorough dependence checking is limited to the changed function of the program under test. A test generation approach that covers the specified function in the program under test at a local high-level coverage is designed. Preliminary experiments show that our approach can reveal most effects of changes with fewer tests.