andygrove opened a new issue, #7133: URL: https://github.com/apache/datafusion-comet/issues/7133
## What is the problem the feature request solves? Review capacity is the main bottleneck for Comet. At the moment there are 189 open pull requests (140 non-draft), and a small number of contributors account for a large share of them. Seven contributors without write access each have four or more non-draft PRs open, and the largest holds 12. Reviewers have to triage this queue, and PRs that arrive in bulk are harder to review well and more likely to go stale. Other DataFusion and related projects have run into the same problem as coding agents make it cheap to open many PRs at once, and have capped the number of open PRs for users without write access: - apache/datafusion [#25973](https://github.com/apache/datafusion/pull/25973) limits users without write access to 3 open, non-draft PRs. It follows the [DataFusion discussion #24413](https://github.com/apache/datafusion/discussions/24413), where most people voted for a limit of 3. - apache/airflow [#73758](https://github.com/apache/airflow/pull/73758) made the same change. ASF Infra implements this as an `.asf.yaml` feature ([apache/infrastructure-asfyaml#111](https://github.com/apache/infrastructure-asfyaml/pull/111)): ```yaml github: pull_requests: creation_cap: enabled: true max_open_pull_requests: 3 ``` ## Describe the potential solution Enable `github.pull_requests.creation_cap` in `.asf.yaml`, and document the limit in the contributor guide (the DataFusion PR adds a short paragraph under "Creating Pull Requests"). Questions to settle in this discussion: 1. Should we adopt a cap at all? 2. What should the limit be? DataFusion chose 3. 3. Should draft PRs count? The DataFusion docs say the limit applies to non-draft PRs. 4. How do we handle regular contributors who are not committers and have legitimate reasons to have more open? Allowlist support through `.asf.yaml` is being added in [apache/infrastructure-asfyaml#135](https://github.com/apache/infrastructure-asfyaml/pull/135) and is still open. Until it lands, a maintainer could close and reopen a PR on the contributor's behalf. 5. How do we treat large work that is split into several small PRs? Our guide asks for small, focused PRs, and a hard cap works against that for people with a long series. The usual answer is to open them one at a time as earlier ones merge, or to combine them with one commit per part. ## Additional context Committers and other users with write access are not affected by the cap, so this only applies to outside contributors. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
