Does the 'Golidlocks Conjecture' Apply to Software Reuse?

摘要

Adopters of corporate software reuse programs face important decisions with respect to the size of components added to the reuse repository. Large components offer substantial savings when reused but limited opportunity for reuse; small components afford greater opportunity for reuse, but with less payoff. This suggests the possibility of an “optimal” component size, where the reuse benefit is at a maximum. In the software engineering discipline, this relationship – termed the Goldilocks Principle - has been empirically observed in software development, software testing, and software maintenance. This paper examines whether this relationship also applies for software reuse. In order to understand the effects of component size and repository size on the benefits of a reuse program this paper extends an empirically grounded reuse model to assess the effects of component size on reuse savings. The study finds that a variant of the Goldilocks Principle applies with respect to both component and repository size, suggesting that uncontrolled growth of a reuse repository and an inappropriate choice of component size may reduce benefits obtained from reuse.