Um estudo de caracterização de mudanças arquiteturais em projetos de software livre

Abstract

Context. The literature background indicates that the scarcity of architectural documentation predominates in free software projects. In this sense, several techniques have been proposed to retrieve information about the architecture; most of them focus on structural aspects from the source code, an element that accompanies the evolution of the software product. Problem. However, these techniques do not recover other information about the architecture (nor its evolution), such as the goals, constraints, and motivations, non-functional requirements, data access mechanisms, or communication mechanisms used in the project. Goal. The purpose of this work is to characterize architectural changes in Free Software projects, presenting the impacts caused by them in the project's source code compared with other types of changes without architectural implication. The hypotheses are that the commits messages record information about changes in the architectural project. This type of change impacts the project code differently compared to changes that affect architecture. Methodology. Initially, we conducted a study where, from repository mining, we identified information about architectural changes contained in commit messages. From the data collected, we conducted an exploratory study that characterized such changes regarding (i) the moment in the project's evolution when they took place, (ii) who their authors are, (iii) which architectural topics are modified, besides (iv) the characteristics of these messages and (v) the group of modules affected by the change. After this study, we conducted a second empirical study involving the extraction of metrics from the source code of the versions of the project identified as having undergone modifications in its architecture. Finally, we conducted a study to evaluate the results obtained, considering the feedback of both developers and researchers in Architecture and Software Engineering regarding the characterization of architectural changes. Results. Among the results, we can highlight that periods encompassing the release of new versions are concentrators of architectural changes; the main collaborators of the project are also the collaborators who modify the architecture the most; all the architectural topics of the project are subject to modifications but at different scales. About the second study, we highlight three results. The size of the code tends to increase. The complexity and cohesion of the code tend to remain constant. The coupling of the code tends to increase for two metrics and to remain constant in three others. Finally, in the third study, we hear developers and researchers about the results of the previous two studies. They confirmed the hypothesis that architectural documentation is scarce in Open Source projects. They claim that the main strategy to retrieve information about architecture is by analyzing the source code in several ways. Most research participants agree with the results obtained by this study about the variation of code metrics. They believe that the variations are generalizable and are recurrent in the projects that they research and develop. Some developers showed slight disagreements from the variations of metrics obtained. Regarding commit messages as sources of information about architecture, most developers and researchers agree that it is a promising source of information, although the ideal scenario would be to combine them with other information from the project.