Date: Tue, 25 Sep 2012 15:31:37 -0400
From: Austin Clements <amdragon@MIT.EDU>
To: Olivier Berger <olivier.berger@it-sudparis.eu>
Subject: Re: Provide an option to make thread summaries keep initial subject
Message-ID: <20120925193137.GM26662@mit.edu>
References: <87zk4eqech.fsf@inf-8657.int-evry.fr>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87zk4eqech.fsf@inf-8657.int-evry.fr>
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: notmuch@notmuchmail.org
Precedence: list

Quoth Olivier Berger on Sep 25 at  6:03 pm:
> Hi.
> 
> Whenever a participant changes the subject in the middle of a thread,
> the summary reported by notmuch search will change.
> 
> However, the result is that some mails tend to "disappear" from search
> results, when (bad) participants reply instead of composing a new mail,
> and change a subject (see
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=688699 for more details
> on that problem, and a discussion). Of course, they're still there, but
> their subject being masked, one may then grep or C-s for a particular
> subject and miss them.
> 
> I think it would be interesting to allow notmuch to potentially keep the
> original subject and not the most recent one for the thread summaries.
> 
> What do you think ?

I think this would be fantastic.  I've proposed unconditionally
showing the earliest subject before and it seems that people who
correspond mostly with those who have good threading etiquette would
prefer this change, but those who correspond with more people who use
'reply' like an address book prefer the current behavior.

Another option, which I'd like to experiment with but haven't found
the time, is to show *all* distinct subjects for matched messages in a
thread (modulo "Re:", etc) in the summary buffer, probably on multiple
lines.  Since most threads only have a single unique subject, they
would appear just as they do now, but it would be clear when someone
(or something, like git) changed the subject mid-thread.  This
approach would be far more robust while retaining good usability, but
it would require more code than just changing our subject-picking
heuristic.