> On Sep 24, 2017, at 8:28 AM, Thomas Weise <t...@apache.org> wrote: > > +1 for closing inactive PRs after documented period of inactivity > (contributor guidelines) > > There is nothing "draconian" or negative about closing a PR, it is a > function that github provides that should be used to improve collaboration. > PR is a review tool, it is not good to have stale or abandoned PRs sitting > as open. When there is no activity on a PR and it is waiting for action by > the contributor (not ready for review), it should be closed and then > re-opened once the contributor was able to move it forward and it becomes > ready for review. > > Thomas
Please refer to my email again, I am not against closing PR if there is inactivity. My issue is with the time period. In reality, most people will create new PRs instead of reopening old ones and the old context/comments will be forgotten and not addressed. > > > > > > > On Sat, Sep 23, 2017 at 4:46 PM, Vlad Rozov <vro...@apache.org> wrote: > >> On 9/23/17 11:07, Pramod Immaneni wrote: >> >>> On Sat, Sep 23, 2017 at 9:11 AM, Vlad Rozov <vro...@apache.org> wrote: >>> >>> Closing PR does not mean that the work submitted for a review is lost. It >>>> is preserved on a contributor fork. All comments and discussion are also >>>> preserved and PR can be re-open if/when a contributor have time to look >>>> into it. >>>> >>>> Closed PRs don't show up in the main list, folks will typically not go >>> into >>> the closed PR list. Items in the main list could catch someone's >>> attention. >>> >> For items in the main list to catch someone's attention it is necessary to >> have some activity on the PR. Why a committer will look into a stale PR >> where an action item is on a contributor side? In case somebody wants to >> look into inactive PR and take the ownership of it, will it be better to >> look into closed and not merged PRs? >> >>> >>> PR can be closed by any committer who was involved into the discussion >>> with >>> >>>> a proper comment. >>>> >>> >>> It is 1 month of inactivity on the PR. Note that PR review is also done >>>> during spare time and recollecting what happened on an inactive PR after >>>> 3 >>>> month or a year requires extra time/effort from a committer and everyone >>>> who were involved into the review. >>>> >>>> Yes, referring to the same. This would not apply in the usual cases but >>> would be the maximum time period in case of inactivity. One month can >>> easily pass if you are distracted with something, it has happened to me a >>> few times for example when something comes up at work. One month seems >>> short, draconian and unnecessary. A longer time period is needed. While it >>> might take some extra effort from the reviewer and/or contributor it does >>> not pose an undue burden, they can quickly come upto speed looking at the >>> PR log. If the PR were closed and reopened it would be the same. Worse if >>> the PR was closed and a new PR were opened for the same code, the old log >>> would probably not show up and then we need additional guidelines to link >>> the old one, so the concerns raised earlier are still addressed. >>> >> If a contributor is distracted with something and does not have time to >> address review comments, will it be better for the contributor to close the >> PR him/herself as a courtesy to reviewers? PR is supposed to be open when a >> code contribution is ready for others to take a look and provide feedback. >> >> There is nothing draconian in closing inactive PR. Closing PR is not an >> irreversible action - all it says is that a contributor needs time to work >> on the PR and is not actively working on it. I feel that 1 month is good >> enough, you think that 3 month is necessary, can we agree on 2 month? >> >> >>> Majority of projects that have hundreds outstanding PRs did not yet move >>> to >>> >>>> gitbox and do not have write access to their github repos. For such >>>> projects closing and opening closed PRs requires additional effort if >>>> possible at all. >>>> >>>> Not many projects are on gitbox so we don't know which way they would go, >>> so you are right in that we cannot assess them based on their current >>> state, even though I do see some examples of this on gitbox, like >>> cloudstack. In any case, we don't have to necessarily follow what other >>> projects do and can determine what works best for us. Normally I would >>> prefer PRs to remain open for longer for reasons above, but a period like >>> 3 >>> months should be fine. >>> >>> Thanks >>> >>> >>> Thank you, >>>> >>>> Vlad >>>> >>>> >>>> On 9/23/17 08:12, Pramod Immaneni wrote: >>>> >>>> I think one month time period is short since people typically contribute >>>>> in >>>>> their spare time. I have seen PRs being worked on with breaks of more >>>>> than >>>>> a month and I have gone back to PRs with that gap as well. Three months >>>>> would be ideal. >>>>> >>>>> How would this be enforced? Would committers do this ad-hoc, whenever >>>>> they >>>>> see some old PRs? Also, this may not be a problem right now and we may >>>>> go >>>>> with more relaxed timelines (like 1 year) as there are other projects >>>>> that >>>>> have hundreds of outstanding PRs. If the PR is open at least someone >>>>> else >>>>> can take it from there, if the original contributor is no longer active >>>>> on >>>>> it. >>>>> >>>>> Thanks >>>>> >>>>> On Fri, Sep 22, 2017 at 6:48 PM, Vlad Rozov <vro...@apache.org> wrote: >>>>> >>>>> I'd suggest that a PR that lack an activity for more than a month is >>>>> >>>>>> closed. Any objections? >>>>>> >>>>>> Thank you, >>>>>> >>>>>> Vlad >>>>>> >>>>>> >>>>>> >>
