Am 29.04.21 um 18:22 schrieb Konstantin Belousov: > On Thu, Apr 29, 2021 at 05:21:22PM +0200, Michael Gmelin wrote: >> >> On Thu, 29 Apr 2021 07:55:58 -0700 >> Cy Schubert <cy.schub...@cschubert.com> wrote: >> >>> In message <20210429162959.16662...@bsd64.grem.de>, Michael Gmelin >>> writes: >>>> >>>> On Thu, 29 Apr 2021 06:53:18 -0700 >>>> Cy Schubert <cy.schub...@cschubert.com> wrote: >>>> >>>>> git restore the files, git rm the files, rm the files, git stash >>>>> push, all don't remove them from the branch following rebase. And >>>>> subsequent rebase to main will fail. main is ok though. >>>>> >>>>> What is it that you did to sync the index? >>>> >>>> Like in the mail thread linked in the git comment, theoretically a >>>> noop: >>>> >>>> git clone https://g...@repo.freebsd.org/src.git >>>> cd src >>>> touch contrib/bc/*vcx* >>>> git commit -a >>>> git push >>>> >>>> It's a problem caused by .gitattributes being checked in after the >>>> *vcx* files were checked in (again, see the email thread and "man >>>> 1 gitattributes"). >>>> >>>> If it was for me, I would remove the .gitattributes eol specs, they >>>> shouldn't be necessary for our purposes.
Sorry, this was my fault. IIRC I had added the files to the commit with "add *" and thus missed .gitattributes, and I did not know about the effect the delayed commit of that missing file would cause. Maybe the best solution is to revert all the commits and then start over with a single commit that includes .gitattributes. The Windows specific files are not relevant for FreeBSD, but I'd rather perform the vendor import and cherry-pick without too many changes to the upstream version. >>> It's probably growing pains. >>> >>> It appears only one branch, probably not surprisingly my most used >>> branch, was the only one affected. I had planned to MFC to stable/13 and 12 this weekend, but will wait for the issue to be settled in the main branch, now. Regards, STefan
OpenPGP_signature
Description: OpenPGP digital signature
