State-Of-The-Practice in Quality Assurance in Java-Based Open Source Software Development
For researchers and practitioners in software engineering, this paper provides a broad empirical understanding of the current state of combined QA practice usage in open source, but the findings are largely descriptive and incremental.
This study investigates how four quality assurance practices (testing, code review, static analysis, and build automation) are used together in 1,454 Java open source projects on GitHub. It finds that projects typically do not apply all practices intensively, with only weak correlations among them, though more mature projects show higher intensity in some practices.
To ensure the quality of software systems, software engineers can make use of a variety of quality assurance approaches, such as software testing, modern code review, automated static analysis, and build automation. Each of these quality assurance practices has been studied in depth in isolation, but there is a clear knowledge gap when it comes to our understanding of how these approaches are being used in conjunction or not. In our study, we broadly investigate whether and how these quality assurance approaches are being used in conjunction in the development of 1,454 popular open source software projects on GitHub. Our study indicates that typically projects do not follow all quality assurance practices together with high intensity. In fact, we only observe weak correlation among some quality assurance practices. In general, our study provides a deeper understanding of how existing quality assurance approaches are currently being used in Java-based open source software development. Besides, we specifically zoomed in on the more mature projects in our dataset, and generally, we observe that more mature projects are more intense in their application of the quality assurance practices, with more focus on their ASAT usage and code reviewing, but no strong change in their CI usage.