[PATCH] Use parsed subject for mboxes

[Stephen Finucane]

On Mon, 2018-04-09 at 09:00 -0400, Veronika Kabatova wrote:
> ----- Original Message -----
> > From: "Daniel Axtens" <dja at axtens.net>
> > To: "Stephen Finucane" <stephen at that.guru>, patchwork at lists.ozlabs.org
> > Sent: Sunday, April 8, 2018 5:21:48 AM
> > Subject: Re: [PATCH] Use parsed subject for mboxes
> > 
> > Stephen Finucane <stephen at that.guru> writes:
> > 
> > > With a recent change, we started using the original subject header
> > > instead of the one we had already cleaned up at the parsing stage.
> > 
> > Does anyone care about the cleaned up header in the mbox? Git strips it
> > all off anyway... Is there any docs on why we did things that way
> > originally?
> > 
> 
> I agree with Daniel and Johannes. For the mboxes, we should use the original
> subject for consistency with the original patch sent. Scripts might get
> confused by the difference, and there might be ones requiring "PATCH" in the
> subject etc. Unless there is an important reason to use parsed subject in the
> mbox, I'd rather revert this and fix the tests. Especially as it's not even
> the mbox tests that fail (which I've run) but the bundles tests using the
> patch's name to assert the patch was added to the bundle (using
> X-Patchwork-Id for those tests would be more appropriate choice).

Yeah, that's fine with me. It doesn't look like git-am or the likes
care and, as Daniel noted earlier in the thread, no one really reads
mboxes by themselves. TBh, I just wanted to fix the tests and this did
look initially like a regression. If someone's happy to pick this up, I
can review/apply.

Stephen

PS: Might want to add a release note about this change in behavior
though.