Nelson, thanks for reaching out and for the suggestions. We're pretty aware of the problem and have taken some steps lately to improve the process. A few highlights:
* We used to have over 1000 PRs open. We have reduced this down to the 200s by closing stale PRs * We now add a "triage" label to any PR coming from the community * After 7 days, we also add a "needs-attention" to community PRs (if they remain un-reviewed) * We have added several committers over the past few months >From my perspective, we are doing fairly well keeping up with reviews. It's not perfect, but it's much better than it used to be. Kafka is experiencing a good bit of community growth lately which means more PRs and more PR reviews needed. We are growing the number of committers as people get involved in the project, but this is a lagging process (by design). [image: image.png] Also consider that 4.0 is being released which is the most significant Kafka release of recent memory. For now, I think we should stay away from auto-assign PRs since it does not really address the main problem (committer review bandwidth). I would like to see how we do with the label based approach over the next few months after 4.0 is out the door. Thanks! David A On Thu, Feb 27, 2025 at 6:22 AM Nelson B. <bachmanity...@gmail.com> wrote: > Hi all, > > I’d like to suggest a revision to our PR review process. Right now, it > seems like the process is to tag people in comments and hope they respond. > But in my experience, this can take painfully long—especially for > infrequent contributors who aren’t part of any community or organization > that actively contributes to Kafka. > > I totally understand that reviewing PRs is a volunteer effort and no one > gets paid for it, so I hope this doesn’t come across as too harsh. My > suggestion is to update the repo settings to automatically assign reviewers > from the list of committers and PMC members, like other open-source > projects do. We could also set up weekly reminders for PRs that haven’t > been reviewed and tag them with a special label so that volunteers with > free cycles can easily find them. Plus, we might introduce a soft policy > asking assigned reviewers to respond within a week if no one else has. > > Thanks, > NB > -- David Arthur
