I think you are missing the point of the "too many reviewers" comment. The way gerrit was set up to interact with architecture domains, requests for gerrit reviews are sent to all the maintainers of the architecture domain that the package resides in. The result is that maintainers are flooded by gerrit with many requests every day for packages they know they have no expertise in or knowledge of so they just bit bucket all email from gerrit. I've heard this comment from many of the reviewers. BRO (Sasha) set this up based on the architecture definition and says it is up to the architects to decide if more discretion is needed in the system. The list of maintainers really needs to be more focused so the review request email just goes to people who care enough about the package to want to review patches to it.
The 2nd problem is that there are several hundred packages that none of the maintainers cares enough about to ever review patches to them. Regards Joel -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Thiago Macieira Sent: Thursday, January 30, 2014 2:54 PM To: Ohly, Patrick Cc: Philippe Coval; [email protected] Subject: Re: [Dev] State of Pending changes , how to enhance workflow ? ( Was Dear Maintainers ) On quinta-feira, 30 de janeiro de 2014 21:54:20, Patrick Ohly wrote: > On Thu, 2014-01-30 at 10:48 -0800, Thiago Macieira wrote: > > Ensuring that everything got reviewed is the responsibility of the > > maintainer. But it's the *interest* of the contributor, so the > > contributor should pay attention to the state of the contribution. > > > > If a contribution goes unreviewed after a while, ping the reviewers. > > That's impractical when the list of reviewers is larger than a handful. > Who should get contacted? Randomly? Chances are high that the ones not > doing any active work on the package will get pinged, annoying them > further and causing additional delays. Just write a new comment saying "ping". Any of the reviewers is capable of approving the change. The question is only whether they feel confident in doing so. > What's your take on those packages which currently have more than ten > reviewers? Having that many reviewers is great! That means more people to look at the changes, to find potential issues, offer advice as well as to provide round-the- clock ability to review things. Having too few reviewers (or none) is a bad thing -- for example, for my changes in QtCore, very often no one can review them, which means I need to seek exceptions. The problem we have in Tizen is not that we have too few reviewers. It's that each reviewer says "there are others, they'll review" and no one ends up doing it. I would have preferred that the Cc list for each contribution have no more than 10 people. More than that, we've got the effect I mentioned. Also, reviewers are encouraged to keep their Gerrit Dashboard *clean*. Review everything that you can review or remove yourself from the Cc list of a change that you don't feel competent to review[*]. [*] Except if you're the maintainer, then you have to ensure it gets reviewed by you or someone else. > > If it > > still doesn't get reviewed, ping the maintainer (write "maintainer > > ping", for example). Sometimes the maintainer will not notice that > > it's the same contribution that no one else reviewed. > > In Tizen 3.0, we have 1276 packages (= packages with a "tizen" branch). > Of those, 655 have no dedicated maintainer. What should be done about > those? A maintainer needs to be found. Any interested and capable reviewer can volunteer to become the maintainer. Failing that, any competent developer. At the very least, we need a gatekeeper that can approve maintenance tasks. If no maintainer is found, we should conclude that the package is abandoned and should be dropped from Tizen. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center _______________________________________________ Dev mailing list [email protected] https://lists.tizen.org/listinfo/dev
