bug#2282: Bug#2282: rmail-summary splits frame into three parts

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Glenn Morris]

(Replying to old bug)

The behaviour described in the original report happens if you make the
Emacs frame wider than `split-width-threshold'. Setting that variable
to nil will prevent it.

I don't know what to do about this. There are lots of pop-to-buffers
in rmailsum.el, and any of them could cause such behaviour.

See the confusion in bug#1806.

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Juri Linkov]

> The behaviour described in the original report happens if you make the
> Emacs frame wider than `split-width-threshold'. Setting that variable
> to nil will prevent it.
>
> I don't know what to do about this. There are lots of pop-to-buffers
> in rmailsum.el, and any of them could cause such behaviour.
>
> See the confusion in bug#1806.

With the current CVS after `M-x rmail-summary' I see

+---------------+---------------+
|               |               |
| RMAIL-summary |     RMAIL     |
|               |               |
+---------------+---------------+
|                               |
|         RMAIL-summary         |
|                               |
+-------------------------------+

I think for RMAIL both following layouts make equal sense
on a wide frame:

+---------------+---------------+
|               |               |
| RMAIL-summary |     RMAIL     |
|               |               |
|               |               |
|               |               |
|               |               |
|               |               |
+---------------+---------------+

and

+-------------------------------+
|                               |
|        RMAIL-summary          |
|                               |
+-------------------------------+
|                               |
|            RMAIL              |
|                               |
+-------------------------------+

The question is what options are necessary to be able to configure RMAIL
window layouts?  Maybe like `gnus-buffer-configuration' in Gnus?

-- 
Juri Linkov
http://www.jurta.org/emacs/

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Stefan Monnier]

> With the current CVS after `M-x rmail-summary' I see

> +---------------+---------------+
> |               |               |
> | RMAIL-summary |     RMAIL     |
> |               |               |
> +---------------+---------------+
> |                               |
> |         RMAIL-summary         |
> |                               |
> +-------------------------------+

This is a bug, please someone fix it.


        Stefan

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Alfred M. Szmidt]

Thanks for doing a follow up on this bug.

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Glenn Morris]

"Alfred M. Szmidt" wrote:

> Thanks for doing a follow up on this bug.

Could you confirm your problem goes away after:

(setq split-width-threshold nil)


I'm tempted to just replace all pop-to-buffers in rmail*.el with

(defun rmail-pop-to-buffer (buffer-or-name &optional other-window norecord)
  (let (split-width-threshold)
    (pop-to-buffer buffer-or-name other-window norecord)))

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[martin rudalics]

 > I'm tempted to just replace all pop-to-buffers in rmail*.el with
 >
 > (defun rmail-pop-to-buffer (buffer-or-name &optional other-window norecord)
 >   (let (split-width-threshold)
 >     (pop-to-buffer buffer-or-name other-window norecord)))

I suppose the following code in `rmail-new-summary' is responsible for
the behavior described earlier.

       (if (and (one-window-p)
	       pop-up-windows
	       (not pop-up-frames))
	  ;; If there is just one window, put the summary on the top.
	  (progn
	    (split-window (selected-window) rmail-summary-window-size)
	    (select-window (next-window (frame-first-window)))
	    (pop-to-buffer rmail-summary-buffer)
	    ;; If pop-to-buffer did not use that window, delete that
	    ;; window.  (This can happen if it uses another frame.)
	    (if (not (eq rmail-summary-buffer
			 (window-buffer (frame-first-window))))
		(delete-other-windows)))
	(pop-to-buffer rmail-summary-buffer))

What is that code supposed to do what a simple `pop-to-buffer' cannot
accomplish?  If it's to display the summary on top of the frame, then
splitting horizontally obviously won't make sense here.

martin

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Stefan Monnier]

>> I'm tempted to just replace all pop-to-buffers in rmail*.el with
>> 
>> (defun rmail-pop-to-buffer (buffer-or-name &optional other-window norecord)
>> (let (split-width-threshold)
>> (pop-to-buffer buffer-or-name other-window norecord)))

> I suppose the following code in `rmail-new-summary' is responsible for
> the behavior described earlier.

>       (if (and (one-window-p)
> 	       pop-up-windows
> 	       (not pop-up-frames))
> 	  ;; If there is just one window, put the summary on the top.
> 	  (progn
> 	    (split-window (selected-window) rmail-summary-window-size)
> 	    (select-window (next-window (frame-first-window)))
> 	    (pop-to-buffer rmail-summary-buffer)
> 	    ;; If pop-to-buffer did not use that window, delete that
> 	    ;; window.  (This can happen if it uses another frame.)
> 	    (if (not (eq rmail-summary-buffer
> 			 (window-buffer (frame-first-window))))
> 		(delete-other-windows)))
> 	(pop-to-buffer rmail-summary-buffer))

> What is that code supposed to do what a simple `pop-to-buffer' cannot
> accomplish?  If it's to display the summary on top of the frame, then
> splitting horizontally obviously won't make sense here.

Hmm... maybe I misunderstood and the problem may indeed not be in
pop-to-buffer, in the end.


        Stefan

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[Stefan Monnier]

> I'm tempted to just replace all pop-to-buffers in rmail*.el with

> (defun rmail-pop-to-buffer (buffer-or-name &optional other-window norecord)
>   (let (split-width-threshold)
>     (pop-to-buffer buffer-or-name other-window norecord)))

No, please don't: the problem is in pop-to-buffer which should never
create 3 windows out of 1.


        Stefan

bug#2282: Bug#2282: rmail-summary splits frame into three parts

[martin rudalics]

 > No, please don't: the problem is in pop-to-buffer which should never
 > create 3 windows out of 1.

Well, we _could_ implement that ;-)

martin