compose-mail-other-window

compose-mail-other-window

[martin rudalics]

Due to the recent change of `compose-mail' `compose-mail-other-window'
will fail to show the window for editing the message.  Removing the
`save-window-excursion' in `compose-mail' should cure this.

Similar misbehaviors have been reported previously by Alfred M. Szmidt
and John ffitch.

Re: compose-mail-other-window

[Leo]

On 2007-03-24, martin rudalics said:

> Due to the recent change of `compose-mail'
> `compose-mail-other-window' will fail to show the window for editing
> the message.  Removing the `save-window-excursion' in `compose-mail'
> should cure this.
>
> Similar misbehaviors have been reported previously by Alfred
> M. Szmidt and John ffitch.

Indeed, compose-mail-other-window is broken in my Emacs 22.0.95 built
on 2007-03-18. It puts the mail buffer in fundamental mode.

-- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

Re: compose-mail-other-window

[Richard Stallman]

    Due to the recent change of `compose-mail' `compose-mail-other-window'
    will fail to show the window for editing the message.  Removing the
    `save-window-excursion' in `compose-mail' should cure this.

This removes the `save-window-excursion'.  Does this work right?

*** simple.el	19 Mar 2007 17:03:02 -0500	1.852
--- simple.el	25 Mar 2007 11:52:34 -0500	
***************
*** 4729,4739 ****
    (let ((function (get mail-user-agent 'composefunc))
  	result-buffer)
      (if switch-function
! 	(save-window-excursion
! 	  (prog1
! 	      (funcall function to subject other-headers continue
! 		       nil yank-action send-actions)
! 	    (funcall switch-function (current-buffer))))
        (funcall function to subject other-headers continue
  	       nil yank-action send-actions))))
  
--- 4729,4738 ----
    (let ((function (get mail-user-agent 'composefunc))
  	result-buffer)
      (if switch-function
! 	(prog1
! 	    (funcall function to subject other-headers continue
! 		     nil yank-action send-actions)
! 	  (funcall switch-function (current-buffer)))
        (funcall function to subject other-headers continue
  	       nil yank-action send-actions))))

Re: compose-mail-other-window

[martin rudalics]

> This removes the `save-window-excursion'.  Does this work right?

Unfortunately it doesn't - I get two windows displying the *mail* buffer
now.  I back Chong's advice to restore the original behavior.

Re: compose-mail-other-window

[Chong Yidong]

martin rudalics <rudalics@gmx.at> writes:

> Due to the recent change of `compose-mail' `compose-mail-other-window'
> will fail to show the window for editing the message.  Removing the
> `save-window-excursion' in `compose-mail' should cure this.
>
> Similar misbehaviors have been reported previously by Alfred M. Szmidt
> and John ffitch.

I think the offending patch should be reverted.  It tries to avoid
passing the SWITCH-FUNCTION argument to sendmail-user-agent-compose;
if I'm not mistaken, this change originated from Michaël Cadilhac's
2007-03-07 complaint to emacs-devel that report-emacs-bug overrides
mail buffers that are already open.  But this behavior was tolerable;
it doesn't justify working around the neat SWITCH-FUNCTION interface
to sendmail-user-agent-compose.

Re: compose-mail-other-window

[Leo]

----- martin rudalics (2007-03-24) wrote:-----

> Due to the recent change of `compose-mail' `compose-mail-other-window'
> will fail to show the window for editing the message.  Removing the
> `save-window-excursion' in `compose-mail' should cure this.
>
> Similar misbehaviors have been reported previously by Alfred M. Szmidt
> and John ffitch.

This bug is not completely fixed.

To reproduce:
  1. emacs -q
  2. (setq mail-user-agent 'message-user-agent)
  3. C-x 5 m

The mail buffer is still in fundamental mode.

I have also tested it with the default value and gnus-user-agent and it
works fine.

-- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

Re: compose-mail-other-window

[Katsumi Yamaoka]

>>>>> In <m2abvsfmtd.fsf@sl392.st-edmunds.cam.ac.uk> Leo wrote:
> ----- martin rudalics (2007-03-24) wrote:-----

>> Due to the recent change of `compose-mail' `compose-mail-other-window'
>> will fail to show the window for editing the message.  Removing the
>> `save-window-excursion' in `compose-mail' should cure this.
>>
>> Similar misbehaviors have been reported previously by Alfred M. Szmidt
>> and John ffitch.

> This bug is not completely fixed.

> To reproduce:
>   1. emacs -q
>   2. (setq mail-user-agent 'message-user-agent)
>   3. C-x 5 m

> The mail buffer is still in fundamental mode.

> I have also tested it with the default value and gnus-user-agent and it
> works fine.

Sorry, that's my bad.

2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>

	* message.el (message-generate-new-buffers):
[...]
	(message-mail): Prefer `switch-function' if it is given;

It requires running `message-mode' further.  I've fixed it in
both the trunk and the v5-10 branch of the Gnus CVS repository.
I think this change is worth merging to not only the Emacs trunk
but also the EMACS_22_BASE branch.

Regards,

P.S.
http://www.jpl.org/ftp/pub/elisp/message-multiple-frames.el.gz
has already been updated.

Re: compose-mail-other-window

[Chong Yidong]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> 2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>
>
> 	* message.el (message-generate-new-buffers):
> [...]
> 	(message-mail): Prefer `switch-function' if it is given;
>
> It requires running `message-mode' further.  I've fixed it in
> both the trunk and the v5-10 branch of the Gnus CVS repository.
> I think this change is worth merging to not only the Emacs trunk
> but also the EMACS_22_BASE branch.

I checked it, and it looks good.  I checked it into the EMACS_22_BASE
branch.  Thanks.

(I assumed the automagic syncing between Gnus CVS and Emacs CVS HEAD
will take care of the trunk; is that right?)

Re: compose-mail-other-window

[Katsumi Yamaoka]

>>>>> In <87wsytekd0.fsf@stupidchicken.com> Chong Yidong wrote:

> Katsumi Yamaoka <yamaoka@jpl.org> writes:

>> 2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>
>>
>> 	* message.el (message-generate-new-buffers):
>> [...]
>> 	(message-mail): Prefer `switch-function' if it is given;
>>
>> It requires running `message-mode' further.  I've fixed it in
>> both the trunk and the v5-10 branch of the Gnus CVS repository.
>> I think this change is worth merging to not only the Emacs trunk
>> but also the EMACS_22_BASE branch.

> I checked it, and it looks good.  I checked it into the EMACS_22_BASE
> branch.  Thanks.

> (I assumed the automagic syncing between Gnus CVS and Emacs CVS HEAD
> will take care of the trunk; is that right?)

Thank you very much for this.  Could you merge another fix that
I made recently to the EMACS_22_BASE branch, too?   The discussion
is here:

http://news.gmane.org/group/gmane.emacs.gnus.general/thread=64667/force_load=t

And this is the patch:

2007-05-24  Katsumi Yamaoka  <yamaoka@jpl.org>

	* message.el (message-narrow-to-headers-or-head): Ignore
	mail-header-separator in the body.

--8<---------------cut here---------------start------------->8---
*** message.el~	Mon May 28 03:03:46 2007
--- message.el	Mon May 28 03:05:06 2007
***************
*** 2225,2238 ****
    (widen)
    (narrow-to-region
     (goto-char (point-min))
!    (cond
!     ((re-search-forward
!       (concat "^" (regexp-quote mail-header-separator) "\n") nil t)
!      (match-beginning 0))
!     ((search-forward "\n\n" nil t)
!      (1- (point)))
!     (t
!      (point-max))))
    (goto-char (point-min)))
  
  (defun message-news-p ()
--- 2225,2236 ----
    (widen)
    (narrow-to-region
     (goto-char (point-min))
!    (if (re-search-forward (concat "\\(\n\\)\n\\|^\\("
! 				  (regexp-quote mail-header-separator)
! 				  "\n\\)")
! 			  nil t)
!        (or (match-end 1) (match-beginning 2))
!      (point-max)))
    (goto-char (point-min)))
  
  (defun message-news-p ()
--8<---------------cut here---------------end--------------->8---

Thanks in advance.

Re: compose-mail-other-window

[Chong Yidong]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> Thank you very much for this.  Could you merge another fix that
> I made recently to the EMACS_22_BASE branch, too?   The discussion
> is here:
>
> http://news.gmane.org/group/gmane.emacs.gnus.general/thread=64667/force_load=t
>
> And this is the patch:

Checked and merged.  Thanks.

Re: compose-mail-other-window

[Katsumi Yamaoka]

>>>>> In <87odk560wy.fsf@stupidchicken.com> Chong Yidong wrote:

>> http://news.gmane.org/group/gmane.emacs.gnus.general/thread=64667/force_load=t

> Checked and merged.  Thanks.

Thanks a lot!

Re: compose-mail-other-window

[Richard Stallman]

    I checked it, and it looks good.  I checked it into the EMACS_22_BASE
    branch.  Thanks.

I'd like the doc fix in rx.el to be checked, and applied if correct.
Also, Martin's table.el change.

Aside from that, I think we are ready for a release, but let's
first have people run it for a couple of days.

Re: compose-mail-other-window

[Leo]

----- Katsumi Yamaoka (2007-05-28) wrote:-----

> Sorry, that's my bad.
>
> 2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>
>
> 	* message.el (message-generate-new-buffers):
> [...]
> 	(message-mail): Prefer `switch-function' if it is given;

Thank you for the fix.

-- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

Re: compose-mail-other-window

[Richard Stallman]

    2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>

	    * message.el (message-generate-new-buffers):
    [...]
	    (message-mail): Prefer `switch-function' if it is given;

    It requires running `message-mode' further.  I've fixed it in
    both the trunk and the v5-10 branch of the Gnus CVS repository.
    I think this change is worth merging to not only the Emacs trunk
    but also the EMACS_22_BASE branch.

Please fix it in EMACS_22_BASE.

Re: compose-mail-other-window

[Katsumi Yamaoka]

>>>>> In <E1HspBN-0001RB-8m@fencepost.gnu.org> Richard Stallman wrote:
>     2007-03-15  Katsumi Yamaoka  <yamaoka@jpl.org>

> 	    * message.el (message-generate-new-buffers):
>     [...]
> 	    (message-mail): Prefer `switch-function' if it is given;

>     It requires running `message-mode' further.  I've fixed it in
>     both the trunk and the v5-10 branch of the Gnus CVS repository.
>     I think this change is worth merging to not only the Emacs trunk
>     but also the EMACS_22_BASE branch.

> Please fix it in EMACS_22_BASE.

It has already been checked by Chong Yidong.  Thanks.