On 28.11.19 15:42, Michal Hocko wrote: > On Thu 28-11-19 09:30:29, Qian Cai wrote: >> >> >>> On Nov 28, 2019, at 9:03 AM, David Hildenbrand <david@xxxxxxxxxx> wrote: >>> >>> That's why we have linux-next and plenty of people playing with it >>> (including you and me for example). >> >> As mentioned, it is an expensive development practice. Once a patch >> was merged into linux-next, it becomes someone else’s problems >> because if nobody flags it as problematic, all it needs is a good eye >> review and some time before it gets merged into mainline eventually. > > I would tend to agree. linux-next shouldn't be considered a low bar > target. Things should be reviewed before showing up there. There are > obviously some exceptions, as always, but it shouldn't be over used. > > I wish MM patches would be applied to mmotm (and linux-next) more > conservatively. I also agree that it should not be used for basic functional/compile tests (I said "It is a way of giving patches *more* testing."). It should not be the only place to test stuff (especially to let somebody else do it). However, sometimes we really have to get additional test coverage via linux-next, especially for weird archs/configurations/setups. ... and if we don't have enough reviewers, it's really hard to get stuff upstream. I wish MM patches would get reviewed more thoroughly. (If we all make a wish, maybe Santa Clause will listen ;) ) -- Thanks, David / dhildenb
