Thanks for pointing this out Yuseok. I am hitting the same problem. I took a look into this, and it seems the same as you described -- GitHub does not allow contributors without write access to reopen their PR if it was closed by a maintainer. It needs write access to reopen.
To avoid adding burden to maintainers, I suggest introducing a small temporary GitHub workflow to deal with it: Contributors can leave a comment as a command in their own PR, and the workflow reopens it automatically after checking the "closed because of open PR limit" label and a few other conditions (the commenter is the author, and the author is still under the 5 open PR limit). I opened a PR for this and would like some opinions on it: https://github.com/apache/airflow/pull/73747 I also considered simply asking contributors to open a new PR from the same branch instead, but that loses labels maintainers already added (e.g. backport labels), and it goes against what the closure comment promised: nothing is lost and PRs can be reopened. I think this gives contributors less friction while staying within the new limit, without bothering maintainers. But I would still like to hear what maintainers think... In case I am missing something. Thanks, Andrew Yuseok Jo <[email protected]> 於 2026年9月26日週六 上午7:07寫道： > Hi Jarek, > > Thanks for driving this. I agree with the direction, and the guide makes > the next steps clear. > > One thing others may hit too. Reopening my closed PRs does not work. The > CLI fails, and the Reopen button is not shown in the web UI, even though > I'm under the limit and the branches are unchanged. My guess is that GitHub > does not let an author without write access reopen a PR closed by someone > with write access, which would affect everyone in the one-time closure. > > Thanks, > Yuseok > > On Sat, Sep 26, 2026 at 5:56 AM Jarek Potiuk <[email protected]> wrote: > > > Hello everyone, > > > > A quick update: https://github.com/apache/airflow/pull/73469 is > > merged, and the limits are now in place. Contributors without write > > access can have at most 5 open PRs at a time. Committers are not > > affected. > > > > I've also run the one-time closure > > (dev/close_prs_over_open_pr_limit.py). The results: > > > > * 210 PRs closed, from 21 contributors. The largest counts were 31, > > 26, 24 and 21 PRs from a single contributor. Most of the rest had 13 > > or fewer, typically around 5-10. > > * Most of them were recent: 110 opened in September, 63 in August. > > * PRs where a maintainer was already engaged stayed open. > > * Every closed PR has a comment explaining why, with a link to the > > guide (contributing-docs/32_open_pull_request_limit.rst). Each also > > has the "closed because of open PR limit" label, so they are easy to > > find: > > > > > https://github.com/apache/airflow/pulls?q=is%3Apr+label%3A%22closed+because+of+open+PR+limit%22 > > > > If a contributor asks about a closed PR, please point them to the > > guide. It explains how to choose which PRs to reopen, up to the limit. > > > > J. > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [email protected] > > For additional commands, e-mail: [email protected] > > > > >
