Should we delay https://github.com/eclipse/tycho/pull/470 to 3.x?
No, we shouldn't, unless it's a breaking change. Anything that's not a "too breaking change" should go to master ASAP, as usual.
I think we can instantly start with a 3.x on master and create a 2.7.x
branch that we only use for bugfixes and final update for 2022-06 (these
should be easily cherry picked to the 3.x branch then or even irrelevant
-100 for starting some branching workflow with cherry-pick and so on here. It's a pain to maintain, build, release... A linear history and single development branch scales much better and we should stick to it unless we really have very strong reasons not to. We can prepare patches for 3.x as pull requests, keep master targeting next release (2.7.0 now, maybe then 2.8, 2.9...) and when we're ready to switch to 3.x, then master becomes 3.x and the only maintained branch. Support for 2.x could still happen -if someone cares about it which I hope won't be the case- on a 2.x branch.