Brant Knudson wrote:
There seems to be a difference of opinion on whether a patch series should be squashed or not when it's backported to the stable branch. What's everyone's opinion on this? Whatever the result is we should update https://wiki.openstack.org/wiki/StableBranch#Proposing_Fixes .
This might affect some of us more than others because I like to split my bug fixes into first a patch that shows the bug and then a patch that fixes it, so I've generally always got 2 patches to backport.
As someone that works on a project that involves backporting changes to branches that have been abandoned by community it's probably easier to backport a squashed patch. And the note for the OSSA is going to have fewer reviews and be less confusing to the non-initiated if there's only 1 commit.
Keeping them separate usually makes it easier to compare with the master patch. One trick with stable branch reviews is that we don't judge the quality of the original patch (this is what the master review by core reviewers is for). We vouch that the same patch has landed in the master branch, and that it is acceptable for the stable branch. IMHO the first part of the task (comparing stable patches with master patches) is facilitated by proposing similarly-organized series... and since the stable review is simpler than the master review, the number of patches is not that much of an issue ? -- Thierry Carrez (ttx)