Re: [PATCH V2] scripts.send-pull-request: Avoid multiple chain headers

[Patrick Ohly <patrick.ohly@intel.com>]

On Mon, 2016-11-28 at 16:35 -0600, Jose Lamego wrote:
> 
> On 11/28/2016 03:34 PM, Patrick Ohly wrote:
> > On Mon, 2016-11-28 at 14:28 -0600, Jose Lamego wrote:
> >> Agree. Please provide feedback about below comments and I will submit a
> >> v3 patch.
> >>
> >> On 11/28/2016 01:47 PM, Patrick Ohly wrote:
> >>> On Mon, 2016-11-28 at 10:23 -0600, Jose Lamego wrote:
> >> More than 1 "In-Reply-To" and "References" message headers are in
> >> violation of rfc2822 [1] and may cause that some email-related
> >> applications do not point to the appropriate root message in a
> >> conversation/series.
> > 
> > Fixing that makes sense. Just add it as reason and the "why" part is
> > covered.
> > 
> >>> And I don't understand why this proposed change has the described
> >>> effect. Does changing the threading parameters change the output of "git
> >>> send-email" and thus indirectly the mail headers of the following
> >>> patches?
> > 
> > The "how" part still isn't clear to me. Perhaps I'm just dumb, but would
> > you bear with me and explain a bit more how changing the sending of the
> > cover letter affects sending of the patches?

I've tried out your proposed change with
  bash -x ../poky/scripts/send-pull-request --to=patrick.ohly@gmx.de -p pull-11827
where pull-11827 is my recent bitbake submission.

The resulting emails are still broken because that one line that you
modify isn't event used. It's under "if [ $AUTO_CL -eq 1 ]" and I am not
using the -a option that enables that behavior.

Even when I use -a, the result is still broken.

The root cause of the problem is that both create-pull-request and
send-pull-request allow git to insert In-Reply-To headers.

"git send-email --help" explicitly warns about that:

        It is up to the user to ensure that no In-Reply-To header already exists when git send-email is asked
        to add it (especially note that git format-patch can be configured to do the threading itself). Failure
        to do so may not produce the expected result in the recipient’s MUA.

>  What I'm doing
> here is to include no reference to any root message at the first call,
> then including a reference at the second call to the very first message
> in the chain, which is either the cover letter or the patch #1.

No, that doesn't work. Whether the first call uses --no-thread or
--no-chain-reply-to has no effect whatsoever, because when "git
send-email" only sends a single email, it doesn't add headers, and the
second call was left unmodified in your patch.

The right fix (tested successfully here) is to use --no-thread in the
second call which sends the sequence of patches. I'll send my  change
for review separately.

-- 
Best Regards, Patrick Ohly

The content of this message is my personal opinion only and although
I am an employee of Intel, the statements I make here in no way
represent Intel's position on the issue, nor am I authorized to speak
on behalf of Intel on this matter.