Studying the Impact of Continuous Delivery Adoption on Bug-Fixing Time in Apache's Open-Source Projects

Abstract

Buggy software impacts people's lives and businesses. Nowadays, a huge portion of a software project's cost is spent on debugging (finding and fixing bugs). Therefore, reducing the time needed to release new software versions free from bugs becomes crucial. Continuous delivery (CD) arises as an alternative to traditional software release engineering by providing the capability to faster and continuously release software to customers through automated pipelines. Previous studies claim that CD adoption leads to a reduction in the software release cycle time, including the time lag to fix reported bugs (bug-fixing time) and apply correction patches in the affected versions. However, there is a lack of empirical evidence supporting (or not) this claim. To fulfill this gap, we conducted an empirical study to evaluate the impact of CD adoption in the bug-fixing time. We study 25 open-source projects comparing the bug-fixing time before and after adopting CD. Our results show that bug-fixing time after CD adoption becomes shorter (with statistical significance) than the bug-fixing time before CD adoption.