Recomendação de revisores de código no desenvolvimento distribuído de software

Abstract

Code review is one of the main software defect mitigation techniques. Intrinsically collaborative, the process involves the analysis of source code modifications by technical peer in adherence to the guidelines of each repository. However, the efficiency of this approach is directly associated with choosing the appropriate technical peers to conduct the scrutiny. Especially in global software development, where members of the development team are geographically scattered with time and availability constraints and there are cultural and technical differences, choosing the reviewer becomes a complex task without the aid of computational tools. The aim of this paper is to present Gitrev, a tool that gathers reviewer recommendation methods for distributed software development that enhance collaboration in the review process. Taking into account the particularities of decentralised production processes, the proposed methods are based on historical interactions between developers, through the design of a collaborative network based on data from GitHub hosted open source repositories. The evaluation of the presented solution is a historic research based on the related literature and objective metrics of efficiency of the reviewers’ participation and instantiated on four major projects available on the platform. The results shows the recommended reviewers enhance interaction and collaboration among team members and adhere significantly to the objectives of the review process.