From: Jani Nikula <jani@nikula.org>
To: Mark Walters <markwalters1009@gmail.com>, notmuch@notmuchmail.org,
	david@tethera.net
Subject: Re: [PATCH 0/4] notmuch reply bugfix & reply to sender only
In-Reply-To: <87hb0924hx.fsf@qmul.ac.uk>
References: <cover.1325794371.git.jani@nikula.org> <87hb0924hx.fsf@qmul.ac.uk>
User-Agent: Notmuch/0.10.2+182~g93862a2 (http://notmuchmail.org) Emacs/23.3.1
	(i686-pc-linux-gnu)
Date: Fri, 06 Jan 2012 14:34:55 +0200
Message-ID: <87fwft80wg.fsf@nikula.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Precedence: list


Hi Mark -

On Thu, 05 Jan 2012 22:01:30 +0000, Mark Walters <markwalters1009@gmail.com> wrote:
> > Bikeshedding topic #1: How about making replying to just the sender the default
> > in "notmuch reply", and having --reply-all option (instead of --no-reply-all)?
> 
> In id:"87pqn5cg4g.fsf@yoom.home.cworth.org" cworth suggested using
> --reply-to=sender vs. --reply-to=all (keeping the latter the default).

That is a sound proposal too, leaving room for extension if such need
might arise.

> However, one more significant choice is what to do on reply-to-sender to
> an email from the user himself. My patch used the following logic: look
> at the reply-to, from , to, cc lines until you find a non-user address
> and use that line for the reply. My recollection is that this is roughly
> what mutt does.

That is a very good point, and one that my patch fails to address.

> (*) I have a version of that patch-set which applies to master if that
> would be useful to anyone, and I recently started writing tests in
> preparation for re-submitting.

Ah, it's old, no wonder I didn't know about it. Looking at the v2 of it
that I found gmane, it looks like the approach is roughly the same.

I think me passing the 'add' parameter makes things a bit more obvious
and explicit, while your use of "g_mime_message_get_all_recipients
(reply) == NULL" might be more robust (including handling reply to
user's own message). Switching to new style argument parsing is probably
something David will insist on, and that's a ready, independent patch in
my set. On emacs side I think my patch is just slightly cleaner, but no
big difference.

How about you post what you have now as a reply to this thread, and let
others be the judge? I really don't mind whether it's you or I who
finishes this as long as we get the feature, and preferrably combining
the best of what we both have. I'm also open to splitting this between
you and me; just let me know what you think.


BR,
Jani.