An empirical study of cycles among classes in Java

Abstract

Advocates of the design principle avoid cyclic dependencies among modules have argued that cycles are detrimental to software quality attributes such as understandability, testability, reusability, buildability and maintainability, yet folklore suggests such cycles are common in real object-oriented systems. In this paper we present the first significant empirical study of cycles among the classes of 78 open-and closed-source Java applications. We find that, of the applications comprising enough classes to support such a cycle, about 45% have a cycle involving at least 100 classes and around 10% have a cycle involving at least 1,000 classes. We present further empirical evidence to support the contention these cycles are not due to intrinsic interdependencies between particular classes in a domain. Finally, we attempt to gauge the strength of connection among the classes in a cycle using the concept of a minimum edge feedback set.

Design principles in architectural evolution: A case study

“…The complete absence of static dependency cycles among plugins, except for a single temporary exception, may point to a possible threat to the validity of our results, because Melton and Tempero report thousands of cycles among Eclipse's classes, some of these cycles involving hundreds of classes [13]. One possibility is that the metadata does not capture all dependencies between plugins.…”

“…Melton argues this would shed light on how effective such principles are to bring about certain system qualities, and it would align academic research better with practical needs. In a separate paper [13], Melton and Tempero check the design principle that dependencies between classes should be acyclic. They measure across a wide range of Java applications the sizes of all cycles present in each application, obtaining a range of different values.…”

Initial survey of Smart Grid activities in the Norwegian energy sector - use cases, industrial challenges and implications for research

“…It introduce a metric termed "depend-on-cycle" that shows that components with this property share some similarities with those directly in cycles. Second, it extends the metric (class reachability set size) proposed in (Melton and Tempero, 2007a), which is a variant of the CCD metric by Lakos . The new metric, named the "Interface Reduction rate for class reachability set size (IRCRSS)", identifies the reduction rate in the reachability set size between a class and its interface.…”

“…Dependency cycles are detrimental to many software quality attributes such as modifiability, reusability, testability and reliability . It inhibits the formation of components to have manageable size and stand-alone properties (Melton and Tempero, 2007a). It is thus crucial to refactor those locations that are particularly defect and change prone.…”

Rethinking Experiments in a Socio-Technical Perspective: The Case of Software Engineering

“…The rest of the papers consists of 12 case studies [20][21][22][23][24][25][26][27][28][29][30][31], 11 reviews [32][33][34][35][36][37][38][39][40][41][42], 11 empirical analyzes [43][44][45][46][47][48][49][50][51][52][53], three comparative analyzes ( [54][55][56]) and two field studies [57,58].…”