Nudge: Accelerating Overdue Pull Requests toward Completion

Pull requests are a key part of the collaborative software development and code review process today. However, pull can also slow down when reviewer(s) or author do not actively engage with request. In this work, we design an end-to-end service, Nudge, for accelerating overdue toward completion by reminding to their requests. First, use models based on effort estimation machine learning predict time given Second, activity detection filter out that may be but which sufficient action is taking place nonetheless. Last, actor identification understand who blocker request nudge appropriate (author reviewer(s)). The novelty Nudge it succeeds in reducing resolution time, while ensuring developers perceive notifications sent as useful, at scale thousands repositories. randomized trial 147 repositories Microsoft, was able reduce 60% 8,500 requests, compared did send notification. Furthermore, receiving resolved 73% these positive. We observed similar results scaling up deployment 8,000 210,000 during full year. This demonstrates Nudge’s ability our qualitative analysis selection indicates areas future research, such dependencies among developer availability into account.