Work

オープンソースソフトウェアプロジェクトにおけるソフトウェア品質保持に関する研究

Year: 2024
Type: dissertation
Abstract: Software quality is an important indicator to ensure that a software system is secure and can be maintained over time by developers.Poor software quality can cause a system to become bloated, too complex to use, or unsafe (i.e., it contains vulnerabilities).For open-source projects, specifically, it is important to conduct software quality maintenance so that new developers can contribute to the project and dependent projects in the open-source ecosystem do not get negatively impacted.One way to ensure that software quality maintenance is conducted is by adopting proper software engineering practices.This includes (but is not limited to) using the right development tools, adopting an efficient release process, integrating software maintenance as part of the standard development process, and much more.In this thesis, we study how software engineering practices can impact software quality maintenance activities by investigating real world practices in OSS (Open-Source Software) projects.More specifically, we aim to investigate if and how do coarse-grained/high-level and fine-grained/low-level software engineering practices affect software maintenance activities.In this context we define fine-grained software engineering practices as matters that affect daily software development activities.This includes what development tools are used daily by developers, or how often developers conduct maintenance activities such as software testing, bug fixing, or performance optimization.Conversely, we consider coarse-grained software engineering practices as matters that are more generic or big picture and which affect the overall development process rather than daily activities.For example, planning the content of each release, or gathering software requirements from stakeholders are both aspects of software engineering that are only conducted when they are required.The first (coarse-grained) software engineering practice we investigated was Abstract ii the release cycle.With modern software requiring constant changes, release cycles have progressively gotten shorter over time.Since core features are usually the primary focus during software development, our goal was to find out if these shorter releases came at the cost of software quality maintenance.For this study, we therefore selected three large OSS projects that have had experience with shortening their release cycles and investigated if their refactoring activities were impacted by the shorter releases.Since refactoring is one of the most common ways to maintain code quality over time, we used refactoring as a proxy metric to measure if software quality maintenance had been affected by the shorter releases.From this investigation, we were able to find out how the change in release cycle affected refactoring activities in software projects.We were also able to provide some recommendations for developers on how to keep up with software maintenance while following shorter release cycles.We continued our investigation of software engineering practices by looking at a fine-grained aspect of software quality maintenance: software testing.In recent years, the OSS ecosystem has been heavily impacted by vulnerabilities spreading throughout the ecosystem and causing billions of dollars in damage.As a result, a vulnerability detection technique called fuzz testing (or fuzzing) has seen a surge in open-source software.In 2022, the Go programming language released native support for fuzz testing.Due to the importance of finding vulnerabilities in open-source software, we investigated if adding fuzz testing into the standard Go library motivated developers to fuzz their software systems.We data mined 11,341 open-source Go projects on GitHub and measured how much fuzz testing is being conducted in OSS projects.We then compared fuzz testing activities with other testing activities to see if fuzz testing was adopted as a standard testing practice.From this investigation, we were able to establish a list of recommendations for developers such as how fuzzing is best used and what functions of their software are critical to fuzz.From there, we decided to investigate deeper and conduct another study on fuzz testing specifically.Since fuzz testing is much more complex to set up and Kyushu
Author: ヌリ, オリビエ
Language: English
Open Access status: green
OpenAlex ID: W7016586297
Cites: 142
Cited by:
Related to: 10