Hi, On Tue, Nov 29, 2011 at 11:20:33PM +0900, Greg KH wrote: > On Tue, Nov 29, 2011 at 01:50:33PM +0200, Felipe Balbi wrote: > > Hi Greg, > > > > A few more fixes for 3.2. Compile tested with allmodconfig and allyesconfig and > > also boot-tested on my x86 desktop with my DWC3 FPGA card. > > > > All patches have been pending on the mailing list for quite a while and they > > should all be fine by now. > > > > The following changes since commit 118205d6b6752e22e19b771771174e6426582311: > > > > USB: linux-cdc-acm.inf: add support for the acm_ms gadget (2011-11-29 09:59:29 +0900) > > > > are available in the git repository at: > > ra.kernel.org:/pub/scm/linux/kernel/git/balbi/usb.git for-greg > > Ick, what's with the "merge" commit at the end of this tree? Not good > at all, that wasn't needed, and only makes things messier. > > So, I just converted this branch into patches, and applied them, please > rebase your tree now, and don't do that again. I just hope we wouldn't have to keep on rebasing branches, it makes things a lot more difficult for me because everytime I have to re-create my master branch and my own history looks messy. Git is very good with merges and that's what most other maintainers are taking in. See Tony's [1] and Arnd's [2] pull requests for instance: it's just a series of merges of other fixes/cleanups/features branches and that's how things should be with git. Whenever I rebase, I loose my commit hashes so I can't make git actually track anything for me. I can't delete a branch safely because git will complain that my branch isn't fully merged; of course, I can git branch -D to force the deletion but that's really not point of using git. Merges aren't bad at all and you should really learn how to deal with them. I feel much better basing all my changes on top of a Linus' tag than on top of your usb-linus or usb-next branches which could contain completely random stuff which hasn't even been on linux-next long enough for me to trust. Granted, my tree isn't on linux-next but that's, again, not the point. Linus has already complained many times because people send pull requests based off of completely untested commits (because it was rebased a few minutes ago on Linus-of-the-day) and I tend to agree with his view: if I test everything on top of 3.2-rc3 and I send a pull request on top of usb-linus, clearly I didn't test what I asking you to pull and I cannot do daily fetch rebase cycles on top of your branches because I will be loosing the cryptograpic safety of git when doing that. If that's really the way you wanna go, there's little point in sending you a pull request at all, I can just combine all patches and send them as email every now and then so you can apply. Still, I would rather go over your rants everytime I send a pull request than being blamed afterwards because my "commits weren't tested" before sending them upstream. ps: sorry for the url, already fixed it. [1] http://marc.info/?l=linux-omap&m=132209259903432&w=2 [2] http://marc.info/?l=linux-kernel&m=132045094212058&w=2 -- balbi
Attachment:
signature.asc
Description: Digital signature
