[Oliver Runge] Re: bug: wrong order of messages in notmuch-show

[David Bremner]

Oliver Runge <oliver.runge at gmail.com> writes:
>
> Hello, David.
>
> I came across this bug report for notmuch when I wanted to report it myself:
> https://www.mail-archive.com/notmuch@notmuchmail.org/msg46633.html
>
> I didn't see an obvious way to respond on the mailing list (tho I now
> subscribed for next time), so I hope a direct response is ok.

I took the liberty of forwarding the response to the list.
>
> You've already reproduced it, but I wanted to give additional
> information. I used a test case in my emails and threw it at git
> bisect. The commit introducing this behaviour is
> 491b1f4b4082bee18418942846ec6508856be7b4, where you fixed the order
> for a loop. Reverting this results in the correct order for threads
> with an unknown Message-ID in In-Reply-To (which is the case in my
> thread as well).

I suspected something along those lines; good to know for sure. I don't
think reverting the commit is an option in general, since the bug it
fixes is much worse than the one it introduces.

> Not having an email with the Message-ID probably is quite a normal
> situation, for instance when people reply to someone and add someone
> else to CC. I guess the right behaviour would be to simply go by date.

Did you try the series id:20180720233746.2844-1-david at tethera.net [1]?
As far as I know it should fix the cases where in-reply-to is missing,
but references is sensible.

>
> I tried to build a test case myself and maybe a patch, but I'm rather
> lost with the tests... and about 200 of them fail for me. No idea
> whether they should pass or not.
>

The tests should definitely all pass (except the 5 or so marked
BROKEN). Perhaps you can start a seperate thread about that. It would be
useful to know the OS, and anything peculiar about your setup.


[1]: https://www.mail-archive.com/notmuch@notmuchmail.org/msg46660.html