Study And Implementation Of Methodology For Identifying Open Source Software Dependency Satisfiability

The development of network technology greatly promoted the expansion and prevalence of Open Source Software (OSS), which enables the shift of software development from the centralized mode to the distributed mode. This transformation significantly improves source code reuse, but also introduces software dependency problems. Although the tools of software management in open source software distribution systems can perform simple dependency identification, there are still no efficient solutions to complex dependencies, and because the softwares update very fast, which lead to instability of the dependancies and the distribution system.This thesis is based on National 863 Project "Research on supporting techniques and applications of a distributed component library service platform (No:2006AA01Z191)", the research theme of which is to study the application of software distribution systems in embedded open source software development. According to the similarity between the functionalities of open source software distribution side and the component library of the service platform, this thesis first analyzes the principles and defects of traditional software distribution systems, and then presents a CNF-SAT solution to identify software dependency satisfiability on the distribution side, which can greatly improve the stability of such systems.This thesis first introduces the fundamental principles of software distribution systems and the characteristics of open source software, and then studies the software dependency problem, especially the problem based on software package management techniques. The problems and limitations of current identification policies are analyzed. Based on these problems, we present a novel method to translate the software dependency satisfiability identification problem into a CNF-SAT problem, and identify the dependency conflicts by solve the SAT problem. A comprehensive set of mapping rules are presented in this thesis and an efficient MiniSAT algorithm is implemented to solve the problems. Experiment results show that this solution can improve the identification of software dependencies and stability of the software distribution system, which is an efficient solution to the software dependency satisfiability identification problem. Future work to deepen the research is given in the last part of this thesis.