A survey of software reuse libraries

The study of storage and retrieval methods of software assets in software libraries gives rise to a number of paradoxes: While this subject has been under investigation for nearly two decades, it still remains an active area of research in software reuse and software engineering; this can be explained by the observation that new technologies (such as the internet, the world wide web, object-oriented programming) keep opening new opportunities for better asset packaging, better library organizations, and larger scale libraries - thereby posing new technical challenges. Also, while many sophisticated solutions have been proposed to this problem, the state of the practice in software reuse is characterized by the use of ad-hoc, low-tech methods; this can be explained by the observation that most existing solutions are either too ineffective to be useful or too intractable to be usable. Finally, while it is difficult to imagine a successful software reuse program without a sophisticated, well-tuned, systematic procedure for software component storage and retrieval, it seems many successful software reuse experiments rely on trivial methods of component storage and retrieval; this can be explained by the observation that, in the current state of the practice, software libraries are not the bottleneck of the software reuse process. This paper presents a survey of methods of storage and retrieval of software assets in software libraries. In addition to a review of existing research efforts, the paper makes two contributions. First, a definition of (presumably) orthogonal attributes of storage and retrieval methods; these attributes are used, in turn, to classify existing methods into six broad classes. Second, a definition of (presumably) orthogonal assessment criteria, which include technical, managerial and human factors; these criteria afford us an exhaustive and uniform basis for assessing and comparing individual methods and classes of methods.