A Comparison of Closed-Source and Open-Source Code Static Measures

Most software engineering empirical studies are based on the analysis of open-source code. The reason is that open-source code is readily available, while usually software development organizations do not give access to their code, not even when the purpose is research and the code itself will not be disclosed. As a consequence, the corpus of empirical knowledge is related almost exclusively to open-source software. This poses a quite important question: do the conclusions we draw from the analysis of open-source code apply to close-source code as well? In this paper, a comparison of open-source and closed-source code is performed, to provide some preliminary answers to the question. Specifically, the goal of the paper is to evaluate whether static code measures from open-source code are similar to those obtained from close-source code. To this end, an empirical study was performed, involving closed-source code from two organizations and open-source code from a few different projects. The most popular static code measures were collected using a commercial tool, and compared. The study shows that open-source code measures appear similar to the measures obtained from industrial closed-source code. However, we must note that the study reported here involved just a few industrial projects’ measures. Therefore, replications of the work presented here would be very useful.