[RFC] parsemail.py: Don't search for patches in replies

[Markus Mayer]

On 9 February 2014 17:20, Jeremy Kerr <jk at ozlabs.org> wrote:
> Hi Markus,
>
>> Make sure we don't attempt to search for a patch in a reply e-mail.
>> There are MUAs out there who leave the quoted e-mail intact without
>> prepending quote characters such as ">" at the beginning of each line.
>
> D'oh!

Yeah, I know.

>> When that happens, parse_patch() thinks the quoted patch is new. The
>> result are multiple database entries containing the same patch (one for
>> each such reply) when one would really expect a consolidated thread
>> containing the entire discussion and only one copy of the patch.
>
> OK, sounds like something that we might be able to address, but I don't
> think this is the correct approach (as others have mentioned too). We
> don't want to drop follow-ups, as they may contain patches that we need
> to track.

I didn't particularly like the approach I sent, either. It's not at
all elegant. But in my case, I need something, as I end up with about
100 duplicate patches otherwise. Making it configurable ("Outlook
mode" vs. "patches in replies" mode) was the best idea I could come up
with. If there's a different way, all the better.

> Perhaps there's some other way we could identify these replies that
> shouldn't be parsed as patches? Could you send an unmodified (mbox-form)
> mail that we can take a look at?

Yes, absolutely. I'll send a conversation your way so you can have a
look at all the mail headers.

Thanks,
-Markus

-- 
Markus Mayer
Broadcom Landing Team