The study of reuse in the context of discovering more productive methods of software development has been popular but generally nonproductive for a long time. The difficulty and elusiveness of reuse in software development has been a major factor in the failure of widespread development and application of software reuse techniques. Issues of software reuse have recently been reviewed in an excellent ACM Computing Surveys paper [1]. In an earlier collection, Biggerstaff and Perlis compiled some of the most influential reuse papers [2]. Furthermore, a preliminary study for an axiomatic framework for reuse is finally available [3].
The lack of metric-based investigations in software reuse has always been recognized. One of the earliest metric analysis works across language-specific projects was cowritten by one of the authors of this paper [4]. This paper presents the techniques, nature, and benefits of reuse in an Ada development environment using a metric-based approach. The authors observed and analyzed the effects of reuse in a single organization over a three-year period. They effectively discuss their work, but the interested specialist might need to contact them to obtain further details, especially data. The paper is well organized into four sections and is readable by nonspecialists. After the introduction, in which key works on reuse are referenced, Section 2 introduces the use of metrics for assessment of the reuse process, and Section3 describes the experiment. Section 4 describes the results. The analysis results are described in three groups: the effect of reuse in the development environment; the effectiveness of Ada generics for reuse; and the organizational impact of reuse.
It is difficult to summarize the results in a few sentences. To obtain the most benefit, the reader should read every sentence and study every graph presented in the paper. A wealth of information is compressed into graphs and concisely discussed by the authors. Nonetheless, in a controlled environment, the authors observed the following:
Most benefit from reuse came from direct reuse without modification.
Ada packages and generics facilitate effective reuse.
Lower error rates were observed.
Reduced development effort was observed.
Development efforts for generics and non-generics were comparable.
Metrics can be used to show various reuse trends.
This study is important. It should be read, studied, and learned from.