GNU bug report logs - #78835
Wrong pop-to-buffer behavior after one display-buffer-in-side-window call

Previous Next

Package: emacs;

Reported by: Dmitry Gutov <dmitry <at> gutov.dev>

Date: Thu, 19 Jun 2025 02:20:02 UTC

Severity: normal

Full log


Message #35 received at 78835 <at> debbugs.gnu.org (full text, mbox):

From: Dmitry Gutov <dmitry <at> gutov.dev>
To: martin rudalics <rudalics <at> gmx.at>, 78835 <at> debbugs.gnu.org
Subject: Re: bug#78835: Wrong pop-to-buffer behavior after one
 display-buffer-in-side-window call
Date: Tue, 15 Jul 2025 05:43:12 +0300
On 14/07/2025 17:53, martin rudalics wrote:
>  > In that particular scenario 'C-x v =' called in the window.el buffer 
> popped up the diff buffer which increased the width of the left window 
> (not sure why).
> 
> You typed C-x v = in a configuration with two side-by-side windows,
> right?

Yes. Ones with seemingly equal width.

>  > And then pressing 'q' in it restored that width.
> 
> If the width changed after displaying the buffer, this is the expected
> behavior.

After displaying and/or after quitting it.

>  > If you could come up with some logging code, I'll be sure to install
>  > it. Or not logging but something to be called when the problem does
>  > happen (if something in the configuration history can trigger it?)
> 
> Try the following - it should size the window back iff you type 'y'.
> 
> diff --git a/lisp/window.el b/lisp/window.el
> index e229562163b..05c33665a8a 100644
> --- a/lisp/window.el
> +++ b/lisp/window.el
> @@ -5457,7 +5457,8 @@ quit-restore-window
>         (when (and (integerp (nth 3 quad))
>            (if (window-combined-p window)
>                        (/= (nth 3 quad) (window-total-height window))
> -                   (/= (nth 3 quad) (window-total-width window))))
> +                   (/= (nth 3 quad) (window-total-width window)))
> +         (y-or-n-p (format "%s ... %s " quit-restore quad)))
>       ;; Try to resize WINDOW to its old height but don't signal an
>       ;; error.
>       (condition-case nil

Yeah, I think it works (the message is pretty long, though).

> Here I can trigger it as follows with emacs -Q: Show window.el in a left
> window and a *Help* buffer in the right window.  The ‘quit-restore’
> parameter of the *Help* window is
> 
> (window window #<window 3 on window.el> #<buffer *Help*>)

Not exactly the same value that I see, but I guess that depends on how 
the Help buffer was created anyway.

> Now I do C-x v = in the left window which displays the diff buffer in
> the right window but does _not_ change its width.  Next I change the
> width of the right window by dragging the mode line divider with the
> mouse, type 'q' in the right window and am asked
> 
> (window window #<window 3 on window.el> *Help*) ... (*Help* 1 #<marker 
> at 1160 in *Help*> 93) (y or n) y
> 
> where the 93 is the previous width which gets restored if I type 'y'.
> What do you get?

I get the same (if the windows windows are equal in the beginning). And 
it's somewhat surprising but also logical at the same time.

OT2H, I also see 'vc-diff' resizing the width (evening the windows) if 
its target window takes up less than half of the frame.

Is that a new behavior? I don't think I've noticed it before. Not sure I 
like it in principle either (my widths are usually either even, or more 
rarely uneven with a specific purpose). It also only happens if the 
window spans the whole height, but not inside a vertical split.




This bug report was last modified 27 days ago.

Previous Next


GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.