To minimize noise, we no longer allow pull requests from contributors unaffiliated with the project or the changes proposed. Specifically, pull requests will be restricted if:
There’s no explanation of changes in the body of the commit, and
The author is not a bot account, and
The author is not the owner or a member of the owning organization, and
The author doesn’t have push access to the head and the source branches
Doesn’t this directly conflict with the idea of “great first issue”? “This would be a great first issue…if you can convince me to give you push access to the repo before I know whether or not you are going to commit good code.”
Or am I misunderstanding? I guess if treated as boolean and, an explanation in the commit body is good enough?
But can’t bots, and drive by PRs then… make boilerplate commit bodies and leave boilerplate explanations? I get that it might make humans behave a bit better, but it likely leads to modifed behavior that is just as bad. “Round of changes” with an explanation of “I made a round of changes.”
I can only speak to my experience working as committer to Pony. We do have a bit of an issue with humans doing drive by, no comment commits. We have no issue with bots doing it. I hear, from talking to other maintainers I know, that the drive by, no comment commits from humans get worse as your project grows larger. I assume the change was made to address this concern/problem that I know some open source maintainers raised with GitHub.
I get that. And, have also seen it happen. But it’s still easily thwarted with crap messages, instead of no messages, which means the problem isn’t solved.
To be clear, I don’t think it’s actually solvable if your goal is allowing everyone to contribute unless you change the social behavior. A better tool, in my mind would be a button that closes the PR with an explanation of the social rules of contribution, and an invitation to resubmit after reviewing them, and making changes.
Given the motivation, I hope that it is enough to have push access to the source branch of the PR — i.e. you can PR your changes, but you need some conditions to PR changes by someone else.
Doesn’t this directly conflict with the idea of “great first issue”? “This would be a great first issue…if you can convince me to give you push access to the repo before I know whether or not you are going to commit good code.”
Or am I misunderstanding? I guess if treated as boolean and, an explanation in the commit body is good enough?
What’s been disallowed if PRs from non-committers with no explanation. Having a commit comment would remove the restriction and allow the PR.
At least that is my understanding.
But can’t bots, and drive by PRs then… make boilerplate commit bodies and leave boilerplate explanations? I get that it might make humans behave a bit better, but it likely leads to modifed behavior that is just as bad. “Round of changes” with an explanation of “I made a round of changes.”
I can only speak to my experience working as committer to Pony. We do have a bit of an issue with humans doing drive by, no comment commits. We have no issue with bots doing it. I hear, from talking to other maintainers I know, that the drive by, no comment commits from humans get worse as your project grows larger. I assume the change was made to address this concern/problem that I know some open source maintainers raised with GitHub.
I get that. And, have also seen it happen. But it’s still easily thwarted with crap messages, instead of no messages, which means the problem isn’t solved.
To be clear, I don’t think it’s actually solvable if your goal is allowing everyone to contribute unless you change the social behavior. A better tool, in my mind would be a button that closes the PR with an explanation of the social rules of contribution, and an invitation to resubmit after reviewing them, and making changes.
Given the motivation, I hope that it is enough to have push access to the source branch of the PR — i.e. you can PR your changes, but you need some conditions to PR changes by someone else.