Software change impact analysis is the process of determining the potential effects, or impacts, of a change to a program. Strategies for impact analysis vary in their approach toward the opposing goals of high precision and low analysis time. Fine-grained techniques, such as slicing, can be used to gain very precise knowledge of a change?s impact, but may be prohibitively expensive. Coarse-grained techniques such as method-level impact analyses sacrifice precision for faster analysis.

In this paper, we present static and dynamic method-level impact analysis algorithms that utilize value propagation information from the source code to increase precision and keep analysis times low. We experimentally compare the results of our analyses with common static and dynamic impact analysis techniques. Our results show that the precision of the common method-level analyses can be improved with very little added overhead.