See past discussions: http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/60150/ , http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/65931/ , http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/76459/ , http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/76640/
Principles
Let's agree on the principles, as a required step to agree on the implementation.
- If a patch is ready for code review, it should be reviewed.
- When a patch has zero reviews, the burden is on the reviewers, no matter how old the patch.
- If the problem is that the repository has been unmaintained for a long time, then the patch submitter could be invited to maintain the repo.
- If the search for new maintainers is futile, then the repo should be marked as inactive and the pending patches marked as abandoned.
- If the problem is that the repository has been unmaintained for a long time, then the patch submitter could be invited to maintain the repo.
- When the patch has +1 or equivalent, the burden is on the reviewers, no matter how old the patch. These patches should get priority in the queue.
- When the patch has -1 or equivalent, the burden is on the authors.
- If the patch doesn't have new activity, after a prudential period (1 month?) a bot should post a reminder.
- If after the automatic reminder there is also inactivity, after a prudential period (2 months?), anybody is entitled to mark the patch as abandoned.
- When a patch had negative reviews in the comments not reflected with a -1, anybody is entitled to mark a -1 on behalf of these comments.
- When a patch has no reviews or +1 and is still waiting for review, it is not cool to force a -1 by throwing it to the Continuous Integration pipeline again, without specific negative feedback about the patch itself
- If the patch looks good but needs rebase, it is cool to leave the good feedback in the comments and force a rebase, even if that results in a -1.
- If the patch doesn't have new activity, after a prudential period (1 month?) a bot should post a reminder.
- When a patch has zero reviews, the burden is on the reviewers, no matter how old the patch.
- If a patch is not ready for code review, it should be marked with [WIP] in the subject.
The concepts of +1 / -1 are specific to Gerrit, but if these principles make sense, we surely can adapt or evolve them for Differential.
See Also:
T129068: Improve code contribution guidelines for patch authors
T78639: How to address the long tail of low priority tasks in active projects (abandoning a rotting changeset = declining a rotting task?)