Assessing Object-Oriented Software Systems Based on Change Impact Simulation

Software changes are inevitable during software evolution. And software change propagation intensely increases the difficulty of software maintenance. In this paper, we regard various change requirements as the combination of a series of “atomic change requirement”. Software modifications, which are used to satisfy the “atomic change requirement”, are considered as modifications of a random selected “initial element” and the ripple effects caused by the modifications. Then we propose a method for assessing the change propagation of object-oriented software based on change impact simulation. Firstly, the method to construct a software change propagation model and related software metric indicators are presented. The rationale of this approach is that different strength of coupling has different probability of change propagation. Secondly, an approach for getting the probability of change propagation setting is provided, which is based on change history obtained from software version repositories and different dependence relationships. Finally, the proposed systematic approach has been evaluated on a multiversion medium sized open source object namely Apache Ant, is a java based build tool, which indicates the simplicity and rationality of our approach.