Code review for newcomers

Abstract

Onboarding is a critical stage in the tenure of software developers with a project, because meaningful contribution requires familiarity with the codebase. Some software teams employ practices, such as mentoring, to help new developers get accustomed faster. Code review, i.e., the manual inspection of code changes, is an opportunity for sharing knowledge and helping with onboarding. In this study, we investigate whether and how contributions from developers with low experience in a project do receive a different treatment during code review. We compare reviewers' experience, metrics of reviewers' attention, and change merge rate between changes from newcomers and from more experienced authors in 60 active open source projects. We find that the only phenomenon that is consistent across the vast majority of projects is a lower merge rate for newcomers' changes.

What makes a code change easier to review: an empirical investigation on code change reviewability

“…Positive reinforcement has been shown to play a role in inducing employees of companies to achieve peak productivity and quality of work [44]. Praising change authors for the positive aspects of their contribution could be beneficial irrespective of the authors' seniority within the project, although this could be even more valuable for newcomers [59]. Furthermore, authors could get a better understanding of what aspects of their code change were considered as good by reviewers, which would lead them to adhere to similar standards with their future contributions.…”

A Systematic Literature Review and Taxonomy of Modern Code Review

“…changes of an author (as a means of measuring familiarity with the project) and the size of the team of reviewers. While two of the studies [46,80] did not find a significant relationship, Lee and Carver [55] indicated a general trend that the number of active reviewers increases when the author's familiarity decreases, suggesting that newcomers receive more attention from invited reviewers. Going in another direction, Yang et al [36] compared active and inactive reviewers of pull-requests.…”

“…Müller [27], McIntosh et al [28], Shimagaki et al [29], Bacchelli and Bird [3], Albayrak and Davenport [30], Beller et al [31], Rigby and Bird [4], German et al [32], Rahman and Roy [33], Sadowski et al [17], Bosu et al [34], Thompson and Wagner [35], Yang et al [36], Kitagawa et al [37], Izquierdo-Cortazar et al [38], Bosu and Carver [39], Spadini et al [40], Baum et al [41], Bird et al [42], Kononenko et al [43], Bosu and Carver [44], Thongtanunam et al [45], Kovalenko and Bacchelli [46], Efstathiou and Spinellis [47], Bosu et al [48], Rahman et al [21], Begel and Vrzakova [49], Dunsmore et al [50], Hirao et al [51], Thongtanunam et al [52], Floyd et al [53], Uwano et al [54], Lee and Carver [55], Meneely et al [56], Bosu et al [57], Sutherland and Venolia [58], Chandrika et al [59], Asundi and Jayant [60], MacLeod et al [61], Ueda et al [62], Armstrong et al [63], Thongtanunam et al [5], Ebert et al [64], Kononenko et al…”

“…Eight studies [80,46,55,18,69,45,63,105] explored factors influencing the team size. As key findings, these studies indicate that both the description length [80] and patch size [18] influence the number of reviewers participating in a review-long patch descriptions and small patch size might increase the likelihood of attracting reviewers.…”

“…Three studies [46,55,80] analyze the relationship between the number of previous contributions, i.e. changes of an author (as a means of measuring familiarity with the project) and the size of the team of reviewers.…”