MORCoRA: Multi-Objective Refactoring Recommendation Considering Review Availability

TL;DR

MORCoRA is a multi-objective search-based approach that recommends refactoring sequences which improve code quality while ensuring reviewers are available and suitable, addressing both technical and review constraints.

Contribution

It introduces a novel multi-objective method that considers review availability alongside code quality and semantics in refactoring recommendations.

Findings

Effectively recommends refactoring sequences that meet quality and review criteria.

Enhances code quality and addresses code smells through recommended refactorings.

Recommends reviewers with high expertise and availability for the suggested refactorings.

Abstract

Background: Search-based refactoring involves searching for a sequence of refactorings to achieve specific objectives. Although a typical objective is improving code quality, a different perspective is also required; the searched sequence must undergo review before being applied and may not be applied if the review fails or is postponed due to no proper reviewers. Aim: Therefore, it is essential to ensure that the searched sequence of refactorings can be reviewed promptly by reviewers who meet two criteria: 1) having enough expertise and 2) being free of heavy workload. The two criteria are regarded as the review availability of the refactoring sequence. Method: We propose MORCoRA, a multi-objective search-based technique that can search for code quality improvable, semantic preserved, and high review availability possessed refactoring sequences and corresponding proper reviewers. Results: We evaluate MORCoRA on six open-source repositories. The quantitative analysis reveals that MORCoRA can effectively recommend refactoring sequences that fit the requirements. The qualitative analysis demonstrates that the refactorings recommended by MORCoRA can enhance code quality and effectively address code smells. Furthermore, the recommended reviewers for those refactorings possess high expertise and are available to review. Conclusions: We recommend that refactoring recommenders consider both the impact on quality improvement and the developer resources required for review when recommending refactorings.