On Fri, Apr 03, 2020 at 12:10:29PM +0300, Dan Carpenter wrote: > On Thu, Apr 02, 2020 at 03:30:34PM +0000, George Spelvin wrote: > > On Thu, Apr 02, 2020 at 11:27:45AM +0300, Dan Carpenter wrote: > > > I don't know how this patch made it through two versions without anyone > > > complaining that this paragraph should be done as a separate patch... > > > > I often fold comment (and spacing/formatting) patches in to a main > > patch, when touching adjacent code anyway and it doesn't cause > > distracting clutter. > > > > This seemed like such a case, which is why I submitted it as one. > > But it's a bit of style thing. > > > > We're super strict in Staging. :P Greg is more strict than I am. Okay, but it's my fault, not his. >> This should have you Signed-off-by. The Reviewed-by is kind of assumed >>> so you can drop that bit. But everyone who touches a patch needs to >>> add their signed off by. >> >> Er... all he did was add "staging: " to the front of the title. >> >> That's not a change to the code at all, and as trivial a change >> to the commit message as adding "Reviewed-by:" to the end. >> We don't need S-o-b for such things or we'd end up in a horrible >> infinite recursion. > > You've misunderstood. He sent the email so he has to add his > Signed-off-by. It's not at all related to changing anything in the > patch. That's how sign offs work. Looking at my commits (just because I remember how they went in), you seem to be right, but damn, submitting-patches.rst could be clearer on the subject. I understand that it's addressed more to patch authors than maintainers forwarding them, but I've read that thing a dozen times, and the description of S-o-b always seemed to be about copyright. So I had assumed that edits which were below the de minimus standard of copyright didn't need a separate S-o-b. Am I right that there should be an S-o-b from everyone from the patch author to the patch committer (as recorded in git)? And the one exception is that we don't need S-o-b for git pulls after that, because the merge commits record the information? For example, my patch series ending at 4684fe95300c (v4.7-rc1~8^2) only has my S-o-b because it was pulled straight from my git server and merge 7e0fb73c52c4 (v4.7-rc1~8) records who merged it. But b5c56e0cdd62 has an S-o-b from both akpm and Linus because it went to akpm, into his quilt, and then as a patch series to Linus, who committed it. All of which is eactly why git-am has a -s option. That's not a hard rule to understand, but I wish submitting-patches *said* so somewhere, rather than having it be implied by the existence of option (c) in the DCO and the fact that it's *doesn't* say that someone else's S-o-b will suffice. And the git merge exception should be stated, because otherwise it's not clear what the limits of that exception are. I had assumed that accumulating and forwarding patches in general was okay without a S-o-b. So thank you for enlightening me, and if you can confirm the rules, I'll prepare a Documentation/ patch to reduce re-occurrence.
