Re: git-publish, --pull-request and --signoff (was: Re: [PULL 0/9] Ide patches)

[Stefan Hajnoczi]

On Tue, Nov 5, 2019 at 9:22 PM Eduardo Habkost <address@hidden> wrote:
> On Tue, Nov 05, 2019 at 09:17:42PM +0100, Stefan Hajnoczi wrote:
> > On Thu, Oct 31, 2019 at 5:07 PM John Snow <address@hidden> wrote:
> > > On 10/31/19 11:02 AM, Peter Maydell wrote:
> > > > On Thu, 31 Oct 2019 at 10:59, John Snow <address@hidden> wrote:
> > > >>
> > > >> The following changes since commit 
> > > >> 68d8ef4ec540682c3538d4963e836e43a211dd17:
> > > >>
> > > >>   Merge remote-tracking branch 
> > > >> 'remotes/stsquad/tags/pull-tcg-plugins-281019-4' into staging 
> > > >> (2019-10-30 14:10:32 +0000)
> > > >>
> > > >> are available in the Git repository at:
> > > >>
> > > >>   https://github.com/jnsnow/qemu.git tags/ide-pull-request
> > > >>
> > > >> for you to fetch changes up to 
> > > >> c35564caf20e8d3431786dddf0fa513daa7d7f3c:
> > > >>
> > > >>   hd-geo-test: Add tests for lchs override (2019-10-31 06:11:34 -0400)
> > > >>
> > > >> ----------------------------------------------------------------
> > > >> Pull request
> > > >>
> > > >
> > > > Hi -- this passed the merge tests but it looks like you forgot
> > > > to add your signed-off by line as the submaintainer to Sam's
> > > > patches. Could you fix that up and resend, please?
> > > >
> > > > thanks
> > > > -- PMM
> > > >
> > >
> > > I bit myself twice with this now: adding --signoff to a pull request
> > > signs the messages that get sent to list, but not the ones that get 
> > > staged.
> > >
> > > Could always be a bug in my local copy, but I'm documenting it on the
> > > list, in case I don't get time to look at this in the next 24h.
> >
> > Do you mean Signed-off-by is only added to emails that are sent and
> > not to the actual commits in your repo?
> >
> > This is how git-format-patch(1) --signoff works.  git-publish does not
> > modify local commits either.
> >
> > Some people would probably be surprised if git-publish modified their
> > commit history.
> >
> > I'm not sure what the best solution here is, aside from introducing a
> > separate signoff option called --apply-signoff or similar so there is
> > no confusion.
>
> I would make git-publish error out if --signoff and
> --pull-request are used simultaneously.  I can't think of a
> justification for having the email contents not match the git tag
> contents in a pull request.

Sounds good!

Stefan