Characterizing refactoring-inducing pull requests.

Abstract

Pull-based development has shaped the practice of Modern Code Review (MCR), in which reviewers can suggest code improvements, such as refactorings, through comments and commits in Pull Requests (PRs). Past MCR studies treat all PRs as similar, regardless of whether they induce refactoring or not. We define a PR as refactoring inducing when refactoring edits are performed after the initial commit(s) as either a result of reviewers’ comments or spontaneous actions carried out by the PR author. This work explores code reviewing-related aspects intending to characterize refactoring inducing PRs. For that, we mined refactoring edits and code review data from GitHub. Then, we carried out empirical studies to identify similarities/dissimilarities between refactoring-inducing and non-refactoring-inducing PRs and characterize code review and refactoring edits in refactoring-inducing PRs. We found significant differences between refactoring-inducing and non-refactoring-inducing PRs and empirical evidence on the relevance of code review to refactoring edits at the PR level. We observed motivating factors behind refactoring-inducing PRs, identified structural aspects of review comments in refactoring-inducing and non-refactoring-inducing PRs, and proposed guidelines for a more productive code review. We also found empirical evidence on technical aspects characterizing refactorings in refactoring-inducing PRs. Our findings suggest directions for researchers, practitioners, and tool builders to improve practices around pull-based code review.