From: Thomas Jost <schnouki@schnouki.net>
To: David Bremner <david@tethera.net>,
	Jameson Graef Rollins <jrollins@finestructure.net>, notmuch@notmuchmail.org
Subject: Re: [PATCH v3 1/4] emacs: Let the user choose where to compose new
	mails
In-Reply-To: <878vhxt5bt.fsf@zancas.localnet>
References: <87d3cx2t38.fsf@rocinante.cs.unb.ca>
	<1323797532-597-1-git-send-email-schnouki@schnouki.net>
	<1323797532-597-2-git-send-email-schnouki@schnouki.net>
	<87pqbayukw.fsf@servo.finestructure.net>
	<878vhxt5bt.fsf@zancas.localnet>
User-Agent: Notmuch/0.12+100~gbed9ca7 (http://notmuchmail.org) Emacs/24.0.95.1
	(x86_64-unknown-linux-gnu)
Date: Mon, 23 Apr 2012 00:25:38 +0200
Message-ID: <87y5pn76al.fsf@schnouki.net>
MIME-Version: 1.0
Content-Type: multipart/signed; boundary="=-=-=";
	micalg=pgp-sha1; protocol="application/pgp-signature"
Precedence: list

--=-=-=
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

Le 15 avril 2012 =C3=A0 16:52 CEST, David Bremner a =C3=A9crit :
> Jameson Graef Rollins <jrollins@finestructure.net> writes:
>
>> I think the issues that David was experiencing have to do with flakiness
>> in emacs's dedicated windows, not in this patch itself.
>
> Thomas,
>
> Did you have a change to investigate this as proposed in
> id:"87zke0aifa.fsf@thor.loria.fr"?

David,

Sorry for the delay. I did investigate a little bit, but I did not try
to write a patch to fix the wrong behaviour in Emacs 23.

AFAICT, Emacs 23 is just buggy in this case. By reading the code of
message-send-and-exit and message-bury [1], here is what happens when
you call message-send-buffer-and-exit with message-kill-buffer-on-exit
set to nil:
=2D message is sent
=2D buffer is buried with burry-buffer
=2D message-bury: if the window is dedicated and its frame not the only
  visible frame, then this frame is deleted

which explains what happens in Emacs 23 both in daemon and non-daemon
mode.

In Emacs 24 [2], here is what happens:
=2D message is sent
=2D message-bury: buffer is buried with bury-buffer

which is (obviously?) correct.

Really, this looks like a bug in Emacs 23 to me. Emacs 24 has been fixed
by Gnus commits [3] and [4] (maybe [3] is enough, I didn't try). Users
of Emacs 23 can probably just use an up-to-date version of Gnus to have
this issue fixed.

So I'm not sure it would make sense to try to come up with a workaround
in my patch, nor if it would be worth it. Maybe just adding a message
suggesting Emacs 23 users to enable message-kill-buffer-on-exit if they
use the Gnus version shipped with Emacs?

Other than that, Jameson's commit [5] is exactly the same as the one in
my tree with a better commit message, so I'm in favor of pulling it.

[1] http://bzr.savannah.gnu.org/lh/emacs/emacs-23/annotate/head:/lisp/gnus/=
message.el
[2] http://bzr.savannah.gnu.org/lh/emacs/emacs-24/annotate/head:/lisp/gnus/=
message.el
[3] http://git.gnus.org/cgit/gnus.git/commit/?id=3D30eb6d24d30bc028fce91a0c=
62044c5dc1fdd90e
[4] http://git.gnus.org/cgit/gnus.git/commit/?id=3De3fc7cb33eb07dd3b48cfc72=
f0cada1f1edbcb85
[5] id:"1334436137-6099-1-git-send-email-jrollins@finestructure.net"

Regards,

=2D-
Thomas/Schnouki

--=-=-=
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iQEcBAEBAgAGBQJPlIVnAAoJEOQWxbmv9eBxUAcIAJtr9akZGKyHqSgLfnwMxmPc
6KWJt9GHzP3o9iXYaXMPvFqHsWsHXAj/iISu11FRVEe12FliNOcj8v2mvuTzMhIj
mM50z91hh+5tPgY6hiSLwaNtd4bTOrecWRBVLuc/uI9z8zS1NE0Mt363uA2YPSr+
9hSZc9TMmcRIQAdwHLLovALFssXF+ijcY7sNRnrhHXXPI5zI5mGObTs+bPYkuOkB
Rj0izfGOxnbJIeJ0dr75tmlZNDAjWc35Jk/s31bdK4Ccp/AT6wVw56uyTMGy9k3l
ySrCe9L/Ttk+p/6xhm7YifEnkUwBzRawOaStLq3Jdnh87pSM/EW1/lGDW5YaQtU=
=YIlX
-----END PGP SIGNATURE-----
--=-=-=--