bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Rahul Martim Juliato]

[-- Attachment #1: Type: text/plain, Size: 1403 bytes --]

Hello there!

I just wanted to report something I found a bit strange.

I would like some keybindings to jump to the next/previous *non-special*
buffer.

With these settings:

 (setq switch-to-prev-buffer-skip-regexp "\*[^*]+\*")
 (setq switch-to-next-buffer-skip-regexp "\*[^*]+\*")

I have the init.el file open, along with the Messages and scratch
buffers.

When I run (switch-to-next-buffer) or (switch-to-previous-buffer), I get
a special buffer. If I run it again, I return to the file, and running
it once more takes me to another special buffer.

If I open another file, like .bashrc, it works as I initially expected.
Now, (switch-to-next-buffer) switches to init.el, and running it again
returns me to init.el back and forth, without switching to any more
special buffers.

It seems that I need to have at least two non-special buffers open. With
only one, the switch function probably (I haven't checked) reaches the
end of the buffer list without finding a match and returns the last
buffer.

Should it work this way? If there are no matches, shouldn't it stay on
the current buffer and/or display a message like "no next/prev buffer"
in the echo area?

Thanks!

-- 
Rahul Martim Juliato
https://www.rahuljuliato.com

PGP Fingerprint: 6B68 4353 84E2 2C7E 5A26  B79A C666 FC94 BD7E A483
PGP Public Key : https://www.rahuljuliato.com/rahul_pub_key.asc

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Eli Zaretskii]

> From: Rahul Martim Juliato <rahuljuliato@gmail.com>
> Date: Sun, 29 Sep 2024 01:25:21 -0300
> 
> I just wanted to report something I found a bit strange.
> 
> I would like some keybindings to jump to the next/previous *non-special*
> buffer.
> 
> With these settings:
> 
>  (setq switch-to-prev-buffer-skip-regexp "\*[^*]+\*")
>  (setq switch-to-next-buffer-skip-regexp "\*[^*]+\*")
> 
> I have the init.el file open, along with the Messages and scratch
> buffers.

Did you forget to double the backslashes?

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Rahul Martim Juliato]

[-- Attachment #1: Type: text/plain, Size: 1217 bytes --]

On Sun, Sep 29, 2024 at 08:39:03AM +0300, Eli Zaretskii wrote:
> > From: Rahul Martim Juliato <rahuljuliato@gmail.com>
> > Date: Sun, 29 Sep 2024 01:25:21 -0300
> > 
> > I just wanted to report something I found a bit strange.
> > 
> > I would like some keybindings to jump to the next/previous *non-special*
> > buffer.
> > 
> > With these settings:
> > 
> >  (setq switch-to-prev-buffer-skip-regexp "\*[^*]+\*")
> >  (setq switch-to-next-buffer-skip-regexp "\*[^*]+\*")
> > 
> > I have the init.el file open, along with the Messages and scratch
> > buffers.
> 
> Did you forget to double the backslashes?

Yep, I forgot it. Strange thing is, it worked without it :/ 

Let me correct it, what I now have is:

  (setq switch-to-prev-buffer-skip-regexp "\\*[^*]+\\*")

And I wrongly added a switch-to-next-buffer-skip-regexp, which does not
exist, the switch-to-prev-buffer-skip-regexp variable does it for both
prev/next functions.

That said, the behaviour is the same as I first described.

-- 
Rahul Martim Juliato
https://www.rahuljuliato.com

PGP Fingerprint: 6B68 4353 84E2 2C7E 5A26  B79A C666 FC94 BD7E A483
PGP Public Key : https://www.rahuljuliato.com/rahul_pub_key.asc

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Eli Zaretskii]

> From: Rahul Martim Juliato <rahuljuliato@gmail.com>
> Date: Sun, 29 Sep 2024 09:29:56 -0300
> Cc: 73545@debbugs.gnu.org
> 
> That said, the behaviour is the same as I first described.

If you mean that Emacs switches to a buffer whose name matches
switch-to-prev-buffer-skip-regexp when there are no other buffers to
show, then this is the intended behavior, I think.  Martin, am I
right?

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[martin rudalics via Bug reports for GNU Emacs, the Swiss army knife of text editors]

 > If you mean that Emacs switches to a buffer whose name matches
 > switch-to-prev-buffer-skip-regexp when there are no other buffers to
 > show, then this is the intended behavior, I think.  Martin, am I
 > right?

I don't know about 'switch-to-prev-buffer-skip-regexp'.  Here I have
set 'switch-to-prev-buffer-skip' to the function

(defun skip-undisplayed-buffers (_window buffer _bury-or-kill)
   "Function for `switch-to-prev-buffer-skip'.
Skip BUFFER if it has never been displayed before."
   (zerop (buffer-local-value 'buffer-display-count buffer)))

(setq switch-to-prev-buffer-skip 'skip-undisplayed-buffers)

Now with emacs -Q evaluate that in *scratch* and do C-x <left>.  Gets me
"No previous buffer" and *scratch* stays.  I think the same should
happen if 'switch-to-prev-buffer-skip-regexp' matches *Messages*.  What
is the value of that variable?

martin

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Rahul Martim Juliato]

[-- Attachment #1: Type: text/plain, Size: 1859 bytes --]

On Sun, Sep 29, 2024 at 05:49:35PM +0200, martin rudalics wrote:
> > What
> > is the value of that variable?
> 
> Found as:
> 
> (setq switch-to-prev-buffer-skip-regexp "\\*[^*]+\\*")
> 
> And the explanation is this snippet from 'switch-to-prev-buffer':
> 
>       (when (and skipped (not (functionp switch-to-prev-buffer-skip)))
>         ;; Show first skipped buffer, unless skip was a function.
> 	(setq new-buffer skipped)
> 	(set-window-buffer-start-and-point window new-buffer)))
> 
> That is, if the regexp excludes a specific buffer but
> 'switch-to-prev-buffer' didn't find another suitable buffer, it switches
> to the first excluded buffer.  The Elisp manual is clear about this:
> 
>      This user option should be either a regular expression or a list of
>      regular expressions.  Buffers whose names match one of those
>      regular expressions will be ignored by ‘switch-to-prev-buffer’ and
>      ‘switch-to-next-buffer’ (except when there’s no other buffer to
>      switch to).
> 
> So if you want to make sure that *Messages* gets always excluded, you
> have to specify a function based on that regexp like
> 
> (defun skip-these-buffers (_window buffer _bury-or-kill)
>   "Function for `switch-to-prev-buffer-skip'."
>   (string-match "\\*[^*]+\\*" (buffer-name buffer)))
> 
> (setq switch-to-prev-buffer-skip 'skip-these-buffers)
> 
> martin

Thanks a lot for the clarification!

I just found a bit strange the default behaviour (I guess I am biased by
my use case) being this way.

The snippet you provided is exactly what I was looking for.

I think this is all for this thread :)

-- 
Rahul Martim Juliato
https://www.rahuljuliato.com

PGP Fingerprint: 6B68 4353 84E2 2C7E 5A26  B79A C666 FC94 BD7E A483
PGP Public Key : https://www.rahuljuliato.com/rahul_pub_key.asc

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[martin rudalics via Bug reports for GNU Emacs, the Swiss army knife of text editors]

 > I just found a bit strange the default behaviour (I guess I am biased by
 > my use case) being this way.

Historical reasons.  'switch-to-prev-buffer-skip' appeared in Emacs 27.1
and at that time I decided that if and only if it is a function it will
make sure that the buffer is never switched to.  More or less to avoid
that peoples shoot themselves in their feet.

'switch-to-prev-buffer-skip-regexp' was added in Emacs 29.1 introducing
the following design error (which I wouldn't propose to exploit).  With
emacs -Q do:

(setq switch-to-prev-buffer-skip-regexp "\\*[^*]+\\*")
(setq switch-to-prev-buffer-skip 'ignore)
C-x <left>

The fact that 'switch-to-prev-buffer-skip' was set to a function that
does nothing makes 'switch-to-prev-buffer-skip-regexp' miraculously
work when there are no other buffers left ...

'switch-to-prev-buffer-skip-regexp' should never have been added in the
first place.  Rather 'switch-to-prev-buffer-skip' should have been given
an optional regexp value.  But now it's too late to change that.

martin

bug#73545: 30.0.91; Strange behaviour on switch-to-next-buffer

[Eli Zaretskii]

> Date: Sun, 29 Sep 2024 17:49:35 +0200
> From: martin rudalics <rudalics@gmx.at>
> Cc: 73545@debbugs.gnu.org
> 
>  > What
>  > is the value of that variable?
> 
> Found as:
> 
> (setq switch-to-prev-buffer-skip-regexp "\\*[^*]+\\*")
> 
> And the explanation is this snippet from 'switch-to-prev-buffer':
> 
>        (when (and skipped (not (functionp switch-to-prev-buffer-skip)))
>          ;; Show first skipped buffer, unless skip was a function.
> 	(setq new-buffer skipped)
> 	(set-window-buffer-start-and-point window new-buffer)))
> 
> That is, if the regexp excludes a specific buffer but
> 'switch-to-prev-buffer' didn't find another suitable buffer, it switches
> to the first excluded buffer.  The Elisp manual is clear about this:
> 
>       This user option should be either a regular expression or a list of
>       regular expressions.  Buffers whose names match one of those
>       regular expressions will be ignored by ‘switch-to-prev-buffer’ and
>       ‘switch-to-next-buffer’ (except when there’s no other buffer to
>       switch to).

Thanks, I've now added this caveat to the variable's doc string, and
I'm therefore closing this bug.