Behavioral Evolution of Design Patterns: Understanding Software Reuse Through the Evolution of Pattern Behavior

Abstract

Design patterns represent a means of communicating reusable solutions to common problems, provided they are implemented and maintained correctly. However, many design pattern instances erode as they age, sacrificing qualities they once provided. Identifying instances of pattern decay, or pattern grime, is valuable because it allows for proactive attempts to extend the longevity and reuse of pattern components. Apart from structural decay, design patterns can exhibit symptoms of behavioral decay. We constructed a taxonomy that characterizes these negative behaviors and designed a case study wherein we measured structural and behavioral grime, as well as pattern quality and size, across pattern evolutions pertaining to four design pattern types. We evaluated the relationships between structural and behavioral grime and found statistically significant cases of strong correlations between specific types of structural and behavioral grime. We identified statistically significant relationships between behavioral grime and quality metrics, as well between behavioral grime and pattern size.