Re: Backporting practices

2020-05-07 Thread Christopher
On Wed, May 6, 2020 at 9:40 AM Szalay-Bekő Máté wrote: > > In general I agree. > > But I am a bit hesitant here: > > > branch-3.6: > > - everything, except new features > > Sometimes it is hard to distinguish "new features" and "improvements". I > think this should be decided on a case-by-case bas

Re: Backporting practices

2020-05-06 Thread Szalay-Bekő Máté
In general I agree. But I am a bit hesitant here: > branch-3.6: > - everything, except new features Sometimes it is hard to distinguish "new features" and "improvements". I think this should be decided on a case-by-case basis. E.g. this recent change seems to be an improvement for me https://iss

Re: Backporting practices

2020-05-06 Thread Enrico Olivelli
Andor, In my opinion it is better to port few patches to released branches. Basically only CVEs and critical bugfixes. The rationale is: - every patch that has been prepared for master branch could not fit cleanly to other branches, even if the cherry-pick succeeds, we could break it - we should en

Backporting practices

2020-05-06 Thread Andor Molnar
Hi! We started a discussion on Slack about what to backport to which branches, because I was missing some commits from 3.5 and 3.6 which I think should have been backported for proper maintenance. These 2 branches are currently the active releases of Apache ZooKeeper, so I think it’s important