threads / discuss / 52663

Patchwork instance "Submitter" attribution and GitGitGadget

Subject: Patchwork instance "Submitter" attribution and GitGitGadget

## tl;dr

6 messages between Jan 18, 2020 and Jan 8, 2021.

replies: 5people: 4as markdown or json

Philippe Blain· Jan 18, 2020, 18:51 UTC · lore
Hi Jonathan,
I’ve noticed that all (I think) patch series sent to the list using Gitgitgadget are attributed to Dscho (Submitter is "Johannes Schindelin via GitGitGadget") on the patchwork instance [1]. I don’t know if you are aware of that or if there’s a way to fix it in patchwork…

Cheers, Philippe.

[1] https://patchwork.kernel.org/project/git/list/?submitter=182351
Johannes Schindelin· Jan 20, 2020, 12:16 UTC · re: Philippe Blain · lore

Re: Patchwork instance "Submitter" attribution and GitGitGadget

Hi Phillippe,
On Sat, 18 Jan 2020, Philippe Blain wrote:
> I’ve noticed that all (I think) patch series sent to the list using
> Gitgitgadget are attributed to Dscho (Submitter is "Johannes Schindelin
> via GitGitGadget") on the patchwork instance [1]. I don’t know if you
> are aware of that or if there’s a way to fix it in patchwork…

Right now, it shows "Lucius Hu via GitGitGadget". Clearly, patchwork uses only the email address as identifier, being unprepared to accept that the same email address might be used by multiple contributors.

Ciao, Dscho

Show 5 quoted lines
>
> Cheers,
> Philippe.
>
> [1] https://patchwork.kernel.org/project/git/list/?submitter=182351
Junio C Hamano· Jan 21, 2020, 17:53 UTC · re: Johannes Schindelin · lore

Re: Patchwork instance "Submitter" attribution and GitGitGadget

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
Show 10 quoted lines
> On Sat, 18 Jan 2020, Philippe Blain wrote:
>
>> I’ve noticed that all (I think) patch series sent to the list using
>> Gitgitgadget are attributed to Dscho (Submitter is "Johannes Schindelin
>> via GitGitGadget") on the patchwork instance [1]. I don’t know if you
>> are aware of that or if there’s a way to fix it in patchwork…
>
> Right now, it shows "Lucius Hu via GitGitGadget". Clearly, patchwork uses
> only the email address as identifier, being unprepared to accept that the
> same email address might be used by multiple contributors.

Would it help to use the "Sender:" header? IIUC right now GGG records its name on "From:" with its e-mail and a human-readable name derived from the author of the ptach, but if it can record the true author on "From:" and leave GGG's name on the "Sender:", would patchwork use the "From:" side of the identity instead?

If that works, it would have an additional benefit of not having to add the in-body "From:" to override the mail-header "From:", to avoid attributing the authorship to GGG.

Johannes Schindelin· Jan 7, 2021, 14:35 UTC · re: Junio C Hamano · lore

Re: Patchwork instance "Submitter" attribution and GitGitGadget

Hi Junio,
On Tue, 21 Jan 2020, Junio C Hamano wrote:
Show 18 quoted lines
> Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
>
> > On Sat, 18 Jan 2020, Philippe Blain wrote:
> >
> >> I’ve noticed that all (I think) patch series sent to the list using
> >> Gitgitgadget are attributed to Dscho (Submitter is "Johannes Schindelin
> >> via GitGitGadget") on the patchwork instance [1]. I don’t know if you
> >> are aware of that or if there’s a way to fix it in patchwork…
> >
> > Right now, it shows "Lucius Hu via GitGitGadget". Clearly, patchwork uses
> > only the email address as identifier, being unprepared to accept that the
> > same email address might be used by multiple contributors.
>
> Would it help to use the "Sender:" header?  IIUC right now GGG
> records its name on "From:" with its e-mail and a human-readable
> name derived from the author of the ptach, but if it can record the
> true author on "From:" and leave GGG's name on the "Sender:", would
> patchwork use the "From:" side of the identity instead?

To tie up that loose end: GMail seemed to insist in my tests on replacing the `From:` header, therefore we cannot implement this `Sender:` idea, not using GMail to deliver the GitGitGadget patch series at least.

Ciao, Dscho

Show 6 quoted lines
> If that works, it would have an additional benefit of not having to
> add the in-body "From:" to override the mail-header "From:", to avoid
> attributing the authorship to GGG.
>
>
>
Konstantin Ryabitsev· Jan 7, 2021, 22:37 UTC · re: Johannes Schindelin · lore

Re: Patchwork instance "Submitter" attribution and GitGitGadget

On Thu, Jan 07, 2021 at 03:35:54PM +0100, Johannes Schindelin wrote:
Show 9 quoted lines
> > Would it help to use the "Sender:" header?  IIUC right now GGG
> > records its name on "From:" with its e-mail and a human-readable
> > name derived from the author of the ptach, but if it can record the
> > true author on "From:" and leave GGG's name on the "Sender:", would
> > patchwork use the "From:" side of the identity instead?
> 
> To tie up that loose end: GMail seemed to insist in my tests on replacing
> the `From:` header, therefore we cannot implement this `Sender:` idea, not
> using GMail to deliver the GitGitGadget patch series at least.

Can you possibly set X-Original-From: in the header? Patchwork almost does the right thing with it, though right now it expects that the From: will be set to the list's email address (which is only correct for google groups).

We have a workaround for the current situation on patchwork.kernel.org [1], but it requires parsing the body of the message instead of just relying on the headers.

> > If that works, it would have an additional benefit of not having to
> > add the in-body "From:" to override the mail-header "From:", to avoid
> > attributing the authorship to GGG.

Using the in-body From: is both good and proper for such service. Trying to put the author's email into the actual From will run afoul of SPF.

-K
[1] https://git.kernel.org/pub/scm/infra/patchwork/procmail.git/tree/git.rc
Johannes Schindelin· Jan 8, 2021, 15:07 UTC · re: Konstantin Ryabitsev · lore

Re: Patchwork instance "Submitter" attribution and GitGitGadget

Hi Konst,
On Thu, 7 Jan 2021, Konstantin Ryabitsev wrote:
Show 12 quoted lines
> On Thu, Jan 07, 2021 at 03:35:54PM +0100, Johannes Schindelin wrote:
> > > Would it help to use the "Sender:" header?  IIUC right now GGG
> > > records its name on "From:" with its e-mail and a human-readable
> > > name derived from the author of the ptach, but if it can record the
> > > true author on "From:" and leave GGG's name on the "Sender:", would
> > > patchwork use the "From:" side of the identity instead?
> >
> > To tie up that loose end: GMail seemed to insist in my tests on replacing
> > the `From:` header, therefore we cannot implement this `Sender:` idea, not
> > using GMail to deliver the GitGitGadget patch series at least.
>
> Can you possibly set X-Original-From: in the header?

Potentially. But unfortunately, I am too short on time to perform such experiments right now. And it is always subject to the question what GMail does with this header in the future. `Sender:` used to work IIRC, but now it doesn't.

Show 14 quoted lines
> Patchwork almost does the right thing with it, though right now it
> expects that the From: will be set to the list's email address (which is
> only correct for google groups).
>
> We have a workaround for the current situation on patchwork.kernel.org [1],
> but it requires parsing the body of the message instead of just relying on the
> headers.
>
> > > If that works, it would have an additional benefit of not having to
> > > add the in-body "From:" to override the mail-header "From:", to avoid
> > > attributing the authorship to GGG.
>
> Using the in-body From: is both good and proper for such service. Trying to
> put the author's email into the actual From will run afoul of SPF.
Indeed.

Therefore, I would prefer to keep doing that in GitGitGadget: send from its regular email address, and add the contributor's info in-body.

Ciao, Johannes

Show 5 quoted lines
>
> -K
>
> [1] https://git.kernel.org/pub/scm/infra/patchwork/procmail.git/tree/git.rc
>

← back to recent threads