On Analyzing the Topology of Commit Histories in Decentralized Version Control Systems

摘要

Empirical analysis of software repositories usually deals with linear histories derived from centralized versioning systems. Decentralized version control systems allow a much richer structure of commit histories, which presents features that are typical of complex graph models. In this paper we bring some evidences of how the very structure of these commit histories carries relevant information about the distributed development process. By means of a novel data structure that we formally define, we analyze the topological characteristics of commit graphs of a sample of GIT projects. Our findings point out the existence of common recurrent structural patterns which identically occur in different projects and can be consider building blocks of distributed collaborative development.