Evaluating software maintenance cost using functional redundancy metrics

Source code copying for reuse (code cloning) is often observed in software implementations. Such code cloning causes difficulty when software functionalities are modified: i.e, cloned codes increase the maintenance cost of software. We aim to estimate the maintenance cost caused by clones. We propose a novel approach, which evaluates influence of cloned codes over the maintenance cost. The basic idea is to measure functional redundancy (FR): a degree of propagation of clone-potential functions. FR is measured as follows: first, we cluster functions in the software according to similarities between them. Second, we make an n-ary weighted tree(FR tree) based on the cluster. Finally, we measure FR by weight of each node in FR-tree. In this paper, we describe the details of our proposal. We also apply the approach to 17K-ELOC C code to demonstrate its effectiveness.