> > To magically close #174, I included "closes #174" in my commit message. > GitHub will then automatically close the pull request noted once that > commit message is merged to master. Unless we're doing changes elsewhere > which remove the need for the original PRs, we won't be able to do this. > > There is some hope on the horizon that we will be able to manage PRs > directly in the future due to the ASF's deployment of GitBox: > > https://gitbox.apache.org/ > > From what I've read in recent threads on the Incubator general@ list, > GitBox will allow us to merge/close/etc. PRs like you'd normally expect. My > understanding is that this is a limited deployment at the moment, but > either I'm wrong about that and we can use it now, or we should be able to > use it in the near future. >
Good to know, though hopefully it won't need to be used all that frequently. > For contributions from the community which have been reviewed, received > feedback, and then fallen silent, we end up in an odd position. People get > busy, people go on vacation, and some of these things take time (even if > apparently simple), so it's always possible that the contributor will > appear, take care of the feedback, and all is well. If that really seems to > not be happening ... > > ... the obvious option would be for us to take over the contribution > ourselves (assuming the feedback on the change deals with fixable problems > with the code itself and not with whether the change is even > necessary/desired at all). I don't relish that idea, since as you've noted, > taking such action would reduce community participation. Contributors > should grow into a pattern of continual, beneficial contributions, > eventually graduating into committership. Stepping in and taking over a > change defeats that. > I also don't like this idea. There seems to be an uptick in the past few weeks in community involvement - other folks forking the code and submitting PRs, and I definitely do not want to discourage that. I also understand that people get busy with other stuff and have a hard time getting back around to cleaning things up - I experience this personally when my day job gets busier and I have less time to churn out code or changes to my code. So, I'm definitely sympathetic to it. Is it worth setting some sort of expectation (policy sounds too formal) about how long these PRs are allowed to languish, how many attempts we make to revive, them, etc., before someone else is allowed/encouraged to grab the code and drive it to completion? And, when I say "languish," I don't mean some expectation about how long it takes to get a PR completed - obviously, as you point out below, there are PRs that have complications that drag them out (the RADIUS one is probably a poster child for that sort of scenaio) - I mean, active attempts by members of the community to engage the original author in communication with no response. I'll also be quick to point out that, if we do step in and take over, I think it needs to be in a way that maintains the credit for the previously written code - commit authorship is maintained, such that one's contribution is acknowledged, even if someone else finishes it up. I believe this should be doable with git - you should be able to fork the branch from the clone where the PR came from. > > Setting aside PRs which are moving slowly for known reasons (being > exceptionally large/unfamiliar, etc.), I think we should take all good > faith action that we can to engage or re-engage the original contributor to > continue working with us to bring their contribution upstream. If pings on > GitHub aren't working, then we clearly need to try something else. I'd > suggest reaching out privately via email to find out what's up, and if > there's anything we can do to help move things along. Once they are > engaged, things are healthy, even if it still takes a while for the PR to > get through. > > If even that doesn't work, and the contributor seems to have truly fallen > off the planet, then we can either take things over or reject the change. > > For the two PRs I mentioned above, I'm happy to take a stab at trying to contact those people directly, see if I can locate those e-mail addresses and get a response from them. I guess the remaining questions in my mind are: - What is the threshold for "fallen off the planet?" At some point, in order to move the code along, (I think) we need to draw a line in the sand and say we're going to merge/fork/modify/close the PR, and I think it would be beneficial if contributors understood that going in. - What should we do if we get a response along the lines of "I'm no longer interested," or "I don't have time," or "You're welcome to finish it for me?" Obviously the later indicates their acceptance of additional help in getting it finished up, but it might be because they feel pressured to accomplish something, which may just be a form of driving people away from contributing. -Nick
