MFC procedure?

[Warner Losh]

On Fri, Dec 11, 2020 at 2:40 PM Kristof Provost <kp at freebsd.org> wrote:

> Hi,
>
> A discussion elsewhere about pre-commit triggers checking commit
> permissions reminded me that I don’t yet fully know how we’ll handle
> MFCs. I”m thinking especially about MFCs of commits initially done by
> someone else. A `git cherry-pick <hash>` will result in an author tag
> with the original author.
>
> Do we want to change the committer name? If so, as I understand it
> that’s relatively simple in Git: cherry pick the change, and run `git
> commit -amend —reset-author’ before pushing.
>
> While on the subject of MFC, I’m similarly not quite clear on wether
> we plan to mirror SVN exactly and rewrite commit messages to include the
> ‘MFC <commit hash>’, or if we’re going to do the more git-y thing
> of just cherry-picking and pushing like that.
>
> Perhaps I overlooked it, but I don’t see any documentation on
> changes/expectations for MFCs in a git world. Would it be helpful if I
> wrote up a proposal for whichever option the transition team prefers?
>

I've not written up the proposed MFC procedure. I plan on doing that after
I get the vendor import stuff I'm working on right now done.

I suspect that we'll mirror much of the SVN process and we'll have a messy
period while we figure out what works best for us.

These are all good questions. I think the answers are "we want to add -x to
the git cherry-pick" to cope with the MFC tag, and we want to retain them
all for squashed MFCs. We'll want to reset the author to conform with local
tradition.

I'll post it to this list when I've written it up, and I'll be sure to CC
you.

Warner