Quality, productivity and economic benefits of software reuse: a review of industrial studies

Reviewer: Chris A Mattmann

The current state of the literature for software reuse is a mixed bag of relevant studies with conflicting conclusions unsupported by quantitative data—claims supported by appropriate data, but lacking uniform dimensions to compare against, and with large gaps in terms of the application of statistically relevant methods. At the same time, it is clear that software reuse has definite tangible benefits (for instance, quality improvement, defect minimization, and productivity). Mohagheghi and Conradi tackle this issue using a scholarly approach of surveying, with as much relevant literature between the years of 1994 and 2005 as they were able to find. The authors employ an investigative approach by examining 11 papers from major software engineering journals and conferences that describe case studies and experiments involving software reuse in industrial settings. A five-question review framework to gather and compare the results of all papers selected was then applied. The framework is based on a statistical/experimental methodology in which the 11 case studies are evaluated based on: the approach and results of software reuse; the metrics selected to evaluate how well software reuse aided in the software project; the quantitative data for software reuse reported; useful theories developed based on the reuse data and findings; and any shortcomings of software reuse described. Much of the paper consists of organized tables that compare the results of the 11 studies surveyed based on further classification of each of the five research questions asked by the authors. This was not incidental: the authors mention that one of the two goals of their study is to provide a means for researchers to quickly sift through the relevant software research data from 1994 to 2005 (the other goal was to assess the state of the literature and understanding of software reuse). The summaries at the end of each subsection in Section 4 are very useful, and more relevant than the wordy text prior to them. The conclusion that the process of experimenting and theorizing about software reuse needs much work is backed sufficiently by the authors’ stated difficulty in trying to classify and compare the experimental data and results provided by the 11 studies. In addition, the research agenda suggested by the authors is highly relevant and should probably be applied more generally to software engineering (SE) as a discipline—it is not limited to software reuse. This is a very readable paper for researchers interested in understanding the state of literature in software reuse. It is presented from an empirical SE and experimental perspective. General readers should focus on the section summaries in Section 4. For those readers interested in a more detailed classification (the empirical “nitty gritty”), the tables in the appendix section provide notes from the authors on each study, and comparisons between them using the five-question framework. Online Computing Reviews Service