Software Similarities: a Case Study with Some General Implications

Generic design aims at achieving non-redundancy by unifying differences among similar program concepts or structures at the design and code levels, for simplification purpose. The extent to which similar program structures deliberately spread through programs, indicates that potentials of generic design may not be fully exploited. While not all such recurring similar program structures are necessarily bad, the evidence abounds that most of them hinder program understanding and contribute much to high maintenance cost. In this paper, we present a study of similarity patterns in the Java Buffer library, JDK 1.5. We show that, given the design goals, many similarity patterns in buffer classes were difficult to avoid refactored or otherwise unified, with conventional OO techniques. As the result, we see classes and methods, recurring many times, in the same or slightly modified form. We trace the roots of the problem and argue about the general nature of the problem’s causes and symptoms.