Does shortening the release cycle affect refactoring activities: A case study of the JDT Core, Platform SWT, and UI projects

Context: Several large-scale companies such as Google and Netflix chose to adopt short release cycles (e.g., rapid releases) in recent years. Although this allows these companies to provide updates and features faster for their users, it also causes developers to have less time to dedicate to development activities other than feature development. Objective: In this paper, we investigate how refactoring activities were impacted by the adoption of shorter releases. Method: We extract all refactorings applied over a period of two years during traditional yearly releases and almost two years during shorter quarterly releases in three Eclipse projects. We then analyze both time periods’ refactoring activities to understand how refactoring activities can be impacted by shortening the release cycles. Results: We observe reduced refactoring activities in one project and a decrease in more complex refactoring operations after shortening the release cycles. We also find that weekly efforts dedicated to refactoring activities was lower across all projects after shortening the release cycles. Conclusion: Shorter releases may impact software development tasks such as refactoring in unintended ways. Not applying specific types of refactoring may also affect the software's quality in the long term. Using this case study and past work on shorter releases, potential short release adopters can now better plan their transition to shorter releases knowing which areas of development may be affected.