On Mon, 13 Nov 2017 17:30:14 +0100, Mark Brown wrote:
On Mon, Nov 13, 2017 at 04:54:13PM +0100, Takashi Iwai wrote:
BTW, one thing Linus requested in the last KS is to avoid merging to a stub branch when you're managing topic branches. If any, just smash the rest branches into one of your branches, instead of merging into the upstream point. Just a reminder for the next time.
I saw the LWN writeup after Vinod mentioned it but what he seemed to be talking about was the practice of doing incremental merges to master each time a topic branch is updated and instead requesting that everything just get merged down at the end which is what I do. Unless he says something directly I'm going to assume my understanding is correct, it's how I send all the stuff I send directly to him so if it was bothering him I guess he'd have said something. I can't see that picking a random topic branch is going to be especially helpful either, it'd end up being one of the fixes branches as they go first which'd quite likely already be in his tree anyway and end up looking like an empty branch based off a non-tag which'd probably upset him more. It's also the mechanism I'm using to merge fixes from him so there is stuff on there relative to the other branches.
If it's an actual problem I'm sure he'll say something.
Yeah, that's why I also took your pull request as is at this time. Let's see :)
My understanding is that it's git-request-pull containing the marginal stuff that annoys him. He usually compares the diffstat between the pull-request and the actual merge result, and it doesn't work when the pull-request already contains the old commits in itself.
Takashi