Why and How Bug Blocking Relations Are Breakable: an Empirical Study on Breakable Blocking Bugs
Hao Ren,Yanhui Li,Lin Chen,Yuming Zhou,Changhai Nie
DOI: https://doi.org/10.1016/j.infsof.2023.107354
IF: 3.9
2024-01-01
Information and Software Technology
Abstract:Blocking bugs prevents other bugs from being fixed, which is difficult to repair and negatively impacts software quality. During software maintenance, developers usually try to break the blocking relationship between blocking and blocked bugs, e.g., propose a temporary fix. However, to our knowledge, no studies have investigated why and how blocking relations between bugs are breakable. In this study, we aim to construct an empirical analysis to explore breakable blocking bugs (BBBs). Specifically, we employ quantitative and qualitative analysis to study these BBBs from two aspects. One is to investigate the characteristics of these bugs, and the other is to explore why and how developers break the blocking relationship between bugs during software maintenance. We build a dataset on five large-scale open-source projects and classify bugs into three types (BBBs, normal blocking bugs, and other bugs) to compare the differences between BBBs and other types of bugs. We observe that BBBs have higher levels of involvement, take longer to fix, and involve more complex source code than other bugs. Moreover, we summarize four reasons blocking relationships between bugs are broken, i.e., partial association (41.87%), serious influence (26.40%), time pressure (19.73%), and flawed blocking (12.21%), and three measures developers adopt to break these blocking relationships, i.e., quick patch for blocking bugs (41.33%), quick patch for blocked bugs (38.67%), and ignore the blocking relation and fix blocked bugs directly (20.00%). Through these analyses, it is meaningful for software maintainers to have a deeper understanding of the characteristics and repair practices of BBBs, which will help solve these BBBs effectively in the future.