bug#13361: 24.2.91; [regression] window point incompatible change

bug#13361: 24.2.91; [regression] window point incompatible change

[Leo Liu]

I have just spent this morning trying to fix a 3rd party library
behaviour change only to find this annoying regression.

To reproduce

0. emacs -q
1. Create two buffers A and B and make them both visible in the frame
2. Insert some text in A and leave point at the end of buffer A
3. Switch to buffer B and eval
    
      (with-current-buffer "A" (goto-char (point-min)))

You will see point in buffer A doesn't change.

Leo

bug#13361: 24.2.91; [regression] window point incompatible change

[Leo Liu]

Sorry not a regression. Sorry for the noise.

It is the same (albeit odd) behaviour in 24.2 too. Any idea why it
behaves so subtly like this?

Thanks,
Leo

bug#13361: 24.2.91; [regression] window point incompatible change

[Glenn Morris]

Leo Liu wrote:

> It is the same (albeit odd) behaviour in 24.2 too. Any idea why it
> behaves so subtly like this?

Are you asking why window-point exists and is not always equal to point?

(with-current-buffer "A"
  (message "%s:%s:%s:%s"
           (point)
           (window-point (get-buffer-window "A"))
           (progn
             (goto-char (point-min))
             (point))
           (window-point (get-buffer-window "A")))
  (sit-for 1)
  (set-window-point (get-buffer-window "A") (point-min)))

bug#13361: 24.2.91; [regression] window point incompatible change

[Leo Liu]

On 2013-01-05 16:36 +0800, Glenn Morris wrote:
> Are you asking why window-point exists and is not always equal to point?

Yeah. For example some packages use with-current-buffer and move point
using goto-char and don't realise it can not be moved if the window is
visible but not current.

Leo

bug#13361: 24.2.91; [regression] window point incompatible change

[martin rudalics]

 > Yeah. For example some packages use with-current-buffer and move point
 > using goto-char and don't realise it can not be moved if the window is
 > visible but not current.

If these packages are part of Emacs, please write an appropriate bug
report.  Otherwise, kindly notify the respective author.

Thanks, martin

bug#13361: 24.2.91; [regression] window point incompatible change

[Glenn Morris]

Leo Liu wrote:

> On 2013-01-05 16:36 +0800, Glenn Morris wrote:
>> Are you asking why window-point exists and is not always equal to point?
>
> Yeah. For example some packages use with-current-buffer and move point
> using goto-char and don't realise it can not be moved if the window is
> visible but not current.

As the example I gave showed, point does move. In any case, I don't see
any bug here.

bug#13361: 24.2.91; [regression] window point incompatible change

[Leo Liu]

On 2013-01-06 03:13 +0800, Glenn Morris wrote:
> As the example I gave showed, point does move. In any case, I don't see
> any bug here.

Point is moved in that context but it is reset to window point
immediately after the window being selected. I think this subtle point
would trip a lot of elisp authors. I have already seen one in
slime-fuzzy.el (3rd party lib).

Leo