Thanks for your feedback Stefan! I totally agree that an important reason for tickets being abandoned is that committers don't have the bandwidth to review these tickets since they are focusing on larger-impact tasks. Missing these contributions is very bad for the project not only because we lose the actual minor fixes and improvements that make the project more stable and usable, but also because we don't value the time the people took to the make contribution by properly reviewing them, so they are not encouraged to return and make new contributions, so we lose in the long run by not attracting new contributors which could in turn increase the project contributor's pool.
I agree that auto-closing tickets by itself will not fix the larger issue mentioned before, but it's just an initial step to "clean the past" so we have an actionable backlog after the 4.0 ticket freeze is over. I believe the addition of the "Triage Needed" state helps a lot to filter valid tickets as they are filled, since in the past any opened ticket would go directly into "open" state and pollute the backlog with invalid tickets that bypassed triaging. My initial objective is to ensure that the "Triage Needed" list is small, and the "Open" tickets can be worked on, which is not the current state but it's pretty simple to achieve with the auto-close proposal. A much harder objective is to ensure that all "Patch Available" tickets are reviewed in a timely manner. However the current backlog of abandoned PA tickets is also large and we need to clean it to "start fresh" so the auto-close proposal would also help here. While I recognize the importance of implementing an strategy to avoid useful contributions from being abandoned and I'm personally committed to improving this, I think this is a much lengthier discussion and wanted to limit this thread to the practical objective of cleaning up the current JIRA state, since I believe this will help regardless of the strategy we choose to address the issue of preventing the tickets from being abandoned in the first place. Em seg., 22 de mar. de 2021 às 06:26, Stefan Miklosovic < stefan.mikloso...@instaclustr.com> escreveu: > Hi Paulo, > > it might seem as me complaining but that is not the case here so just > hear me out please. > > I think that the primary reason for tickets not being resolved / they > are abandoned / they are not worked on anymore is not the fact that > they are just "forgotten" but I believe that it is also because after > spending some time around the project, one just feels what ticket has > any reasonable chance to get through and be eventually committed or if > that ticket is just not too important after all (from time-management > point of view) and it just does not make sense to work on it because > the amount of time spent on it to chase committers to look at it is > just not worth it which is quite sad but understandable. I know about > a couple of tickets I would like to see merged but I just do not > bother because I am not a committer myself and begging people for a > couple weeks to look at some minor stuff is just a waste of time for > everybody. But it does not mean that such a ticket is "useless". It is > more about people not having any more bandwidth to look at them and so > on ... > > Minor or low-hanging fruit tickets are easy to work on and they are > usually the entry point for new contributors but those who do have > commit rights are not too often impressed and it stays under the > radar, it is a chicken-egg problem and eventually the project as a > whole loses, contributors are often discouraged and minor fixes sum > over time to quite a lot of changes which makes the difference but > they are not there ... > > Hence while I guess some solutions you propose are sound, it still > does not prevent valid tickets to be abandoned and forgotten > completely. I would yet like to see what mechanism would be put in > action to decide what ticket is relevant or not after being not worked > on for a substantial amount of time. > > Regards > > On Sun, 21 Mar 2021 at 23:06, Paulo Motta <pauloricard...@gmail.com> > wrote: > > > > Hi everyone, > > > > While triaging tickets for Google Summer of Code I went through dozens of > > invalid and stale tickets on JIRA that were still open. In addition to > > other issues, this makes it harder for new contributors to identify valid > > tickets to work on. Ideally any ticket that was triaged by a contributor > > should be good to work on, but this is not the current state of our JIRA. > > > > I was wondering if we should adopt some strategy to auto-close JIRA > tickets > > and/or GitHub PRs after a period of inactivity and wanted to bring this > > discussion to the community. One downside of this is that valid tickets > > might be auto-closed, but we could use a few strategies to mitigate that: > > a) use a long enough period (ie. 2 years), if an issue wasn't worked on > > within this period then it's probably not relevant enough and can be > > abandoned. This would already get rid of a large backlog of abandoned > > tickets dating back to as early as 2014 (or even earlier). > > b) if the issue hasn't been worked on for a large period but the > community > > still finds it relevant, add some tag to prevent that ticket from being > > auto-closed. > > > > I think that this is something that is pretty easy to fix and will make > our > > ticket tracking more reliable and easier for new contributors to get > > started, especially in the post 4.0 world. > > > > Please let me know what do you think, > > > > Paulo > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@cassandra.apache.org > For additional commands, e-mail: dev-h...@cassandra.apache.org > >
