Is It Safe to Uplift This Patch? An Empirical Study on Mozilla Firefox

摘要

In rapid release development processes, patches that fix critical issues, orimplement high-value features are often promoted directly from the developmentchannel to a stabilization channel, potentially skipping one or morestabilization channels. This practice is called patch uplift. Patch uplift isrisky, because patches that are rushed through the stabilization phase can endup introducing regressions in the code. This paper examines patch upliftoperations at Mozilla, with the aim to identify the characteristics of upliftedpatches that introduce regressions. Through statistical and manual analyses, wequantitatively and qualitatively investigate the reasons behind patch upliftdecisions and the characteristics of uplifted patches that introducedregressions. Additionally, we interviewed three Mozilla release managers tounderstand organizational factors that affect patch uplift decisions andoutcomes. Results show that most patches are uplifted because of a wrongfunctionality or a crash. Uplifted patches that lead to faults tend to havelarger patch size, and most of the faults are due to semantic or memory errorsin the patches. Also, release managers are more inclined to accept patch upliftrequests that concern certain specific components, and-or that are submitted bycertain specific developers.