On Fri, 16 Dec 2011 15:45:26 -0800, Jameson Graef Rollins <jrollins@finestructure.net> wrote: > Hey, David. What exactly is the problem here? These seems like it's > actually reasonable behavior when you're using emacs in daemon mode, > yes? I don't actually use notmuch with emacs daemon/client, so I'm not > sure what behavior is expected. The frame started by "emacsclient -c" is no different than a frame started with "emacs". It closes when you type C-x C-c. In particular killing an emacs window should not kill the frame. With this patch, and notmuch-mua-compose-in set to new-window, notmuch does this. > If this is an emacsclient issue, should that prevent the patch from > going through? It is pretty disruptive to have a frame closed unexpectedly, and I think enough people use 'emacsclient -c' that this is not a use case we can dismiss. In any event, I just checked with 'emacs -q', no emacsclient, and the behaviour of 'new-window is strange there too, since it buries the notmuch window, and leaves the window containing the sent message displayed. d