On Jan 27, 2024, at 5:04 PM, Markus Karg <markus@xxxxxxxxxxxxxxx> wrote:
We maybe can simplify that by boiling it down to effectively restricting all limitations of our committer conventions to solely the main branch itself, while enforcing that any binary publications (aka releases and pre-releases) must ONLY happen from the main branch?
I agree (almost). Where I part with you is on “pre-releases”. I think that “release-x.x” branches should not be moved to main/master until the release has been approved. That implies that CI exists that passes a TCK and verifies all expected content documented in the Spec/APIs. For EE11 we would need to be ready for that by 03-29-24. Of course after that time there will likely still be bug fixes, TCK challenges, etc., but those would be made to the master/main branch.
That is why I defined a “working branch” as one that was not associated with a documented release plan. Even if we are not yet working on master/main, once a release plan exists we may want to have a bit more control. Maybe not as restrictive as we do for master/main, but perhaps at least a couple reviews by committers and maybe a 3 business day wait period (unless it is fast-tracked). |