I'm currently scrolling through the open issues, trying to close all that are done, and I stumbled upon #4644, fixed by #4875, which I had hoped would be backported/released/not-kept-in-limbo. I also shied away from labelling it myself, hence it fell through the cracks, and missed the v0.27.8 release brouhaha, hence it shipped without that fix, which would have made me more comfortable about those weird OpenSSL failures.

Hence, I'd like to discuss how should to best manage the labels for releases. Right now, @pks-t started the trend of the backport-$version labels, which I find useful, but I don't think they would scale, and there's a next-release label, which @ethomson used for the recent changelog, but is much more ancient.

Proposal

• I don't think we need the full version in the backport labels, so I'm tempted to switch to MAJOR.MINOR. This limits us, in that we loose (unless you have crazy git skills) the ability to tell at a glance whether a given release has the given fix, but keeps the amount of labels down (we're still v0, and there's already 5 of them), which is what I'm afraid won't scale (I envision a world with a v1.0, v1.1 and v1.2 maintenance branches + master/v1.3, that might be a lot of labels).

• Any merged PR on master gets the next-release label (unless it's targeting a maintenance branch, in which case it gets a backport), and we can discuss if it's eligible for a backport (and where) at that time, and the PR gets tagged as such.

• When preparing a release, you have all the labels needed to know what's going in. For a maintenance release, look at backport, for a new branchpoint, look at next-release. The release is done, unlabel all those we used, and we can start over.

Does that feel like a plan ?

Some additional notes :

I'm unable to do some of the crazier things you can do with git, like identifying whether a given commit has appeared in another branch as a backport, so if you can enlighten be I'd be glad.

I'm unsure what was the original motivation behind the full backport labels, so maybe this is contradicting some meaning I'm unaware of (hence, this discussion).