Patchwork to require git-send-email formatted patches?

[Peter Hutterer]

On Thu, Nov 12, 2015 at 02:20:01PM +0200, Pekka Paalanen wrote:
> On Thu, 12 Nov 2015 11:59:17 +0000
> Damien Lespiau <damien.lespiau at intel.com> wrote:
> 
> > On Wed, Nov 11, 2015 at 02:01:38PM +0200, Pekka Paalanen wrote:
> > > On Tue, 10 Nov 2015 11:35:35 +0000
> > > Damien Lespiau <damien.lespiau at intel.com> wrote:
> > > 
> > > > Something else I noticed on this thead:
> > > > 
> > > > In the reply 20151019013047.GA8175 at jelly.local from Peter, there's a
> > > > diff put there for reference, picked up as a patch by patchwork:
> > > > 
> > > >   http://patchwork.freedesktop.org/patch/msgid/20151019013047.GA8175@jelly.local
> > > > 
> > > > If you want patchwork to not consider those kind of diffs as real
> > > > patches, there's a per-project option I can activate: only consider git
> > > > send-email mails as potential patches.
> > > 
> > > Hi all,
> > > 
> > > I think we could turn that on. What do others think?
> > > 
> > > How is a git-send-email patch recognized? Would we miss patches that
> > > are formatted with git-format-patch but sent by other means than
> > > git-send-email?
> > 
> > git send-email patches are recognized by their X-Mailer header:
> > 
> >   is_git_send_email = mail.get('X-Mailer', '').startswith('git-send-email')
> > 
> > It does mean any diff inlined in an email will be skipped if sent
> > "manually", even it they actually are from git format-patch. I took some
> > care to not cull git format-patch files sent as attachments though, so
> > those should still work with that option enabled.
> 
> Hi,
> 
> looks like Emre's emails do not even have X-Mailer header, and if he
> had, it wouldn't be git. So I don't think we can enable this filtering
> after all.
> 
> I suppose there are also other people who for one reason or another
> don't use git-send-email.
> 
> IMHO, it is less of a burden to prune accidental patches from Patchwork
> than cause people grief by rejecting legitimate patches. Or does that
> screw up the patch revision or series tracking?

fwiw, I use mutt -H for single patches, it's faster than git-send-email
which I use for patch series only.

Cheers,
   Peter