Location of mouse click

Location of mouse click

[Yasushi SHOJI]

[-- Attachment #1: Type: text/plain, Size: 829 bytes --]

Hi,

Emacs seems to disagree with the mouse clicked location of and buffer
lines when there are lines with different height.

Let's say I have short lines with higher height characters and a
longer line with shorter height in a buffer.  The buffer is set
truncate-lines t.  The longer lines, say, have a Japanese character あ.
Mouse clicks work while あs are displayed.  But once they are gone due
to horizontal scrolling to right, the long line moves up.  Now, a
mouse click on the longer line make the buffer scroll back to left,
just like I clicked the line above the long line (but I didn't).

I've tested under both X11 and Wayland session.

I wasn't sure I can explain this in sentences, so I made a
screen-cast.  apologize for a MB attachment.

Is this known behavior?

Best,
--
            yashi

[-- Attachment #2: Screencast from 04-13-2022 09:24:10 PM.webm --]
[-- Type: video/webm, Size: 845946 bytes --]

Re: Location of mouse click

[Eli Zaretskii]

> From: Yasushi SHOJI <yasushi.shoji@gmail.com>
> Date: Wed, 13 Apr 2022 21:44:33 +0900
> 
> Emacs seems to disagree with the mouse clicked location of and buffer
> lines when there are lines with different height.
> 
> Let's say I have short lines with higher height characters and a
> longer line with shorter height in a buffer.  The buffer is set
> truncate-lines t.  The longer lines, say, have a Japanese character あ.
> Mouse clicks work while あs are displayed.  But once they are gone due
> to horizontal scrolling to right, the long line moves up.  Now, a
> mouse click on the longer line make the buffer scroll back to left,
> just like I clicked the line above the long line (but I didn't).
> 
> I've tested under both X11 and Wayland session.
> 
> I wasn't sure I can explain this in sentences, so I made a
> screen-cast.  apologize for a MB attachment.
> 
> Is this known behavior?

Not sure.

Please post a file that can be visited to reproduce the problem, and
please accompany that file with instruction for reproducing starting
from "emacs -Q".  If it is important to click the mouse at specific
places, please try to describe those places as accurately as you can,
to avoid misunderstanding.

Also, please tell which version of Emacs is that.;

Thanks.

Re: Location of mouse click

[Yasushi SHOJI]

[-- Attachment #1: Type: text/plain, Size: 783 bytes --]

Hi,

On Wed, Apr 13, 2022 at 10:30 PM Eli Zaretskii <eliz@gnu.org> wrote:
> Please post a file that can be visited to reproduce the problem, and

Attaching a.txt

> please accompany that file with instruction for reproducing starting
> from "emacs -Q".

0. emacs -Q
1. open a.txt
2. (toggle-truncate-lines)
3. goto the end of the last line
4. make the frame narrow so that "あ"s are not shown
5. click on the last word "dog"

But, this assumes that Japanese and ASCII chars have different heights.

> If it is important to click the mouse at specific

You can click anywhere on the last line.

> Also, please tell which version of Emacs is that.;

- 28.1.50 emacs-28.1-25-g5ee959aa8
- 29.0.50 emacs-28.1-154973-gfbf27afb6a

Best,
-- 
           yashi

[-- Attachment #2: a.txt --]
[-- Type: text/plain, Size: 493 bytes --]

a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a line with slightly higher height あ
a long line. The quick brown fox jumps over the lazy dog. The quick brown fox jumps over the lazy dog.

Re: Location of mouse click

[Eli Zaretskii]

> From: Yasushi SHOJI <yasushi.shoji@gmail.com>
> Date: Wed, 13 Apr 2022 23:51:18 +0900
> Cc: emacs-devel@gnu.org
> 
> On Wed, Apr 13, 2022 at 10:30 PM Eli Zaretskii <eliz@gnu.org> wrote:
> > Please post a file that can be visited to reproduce the problem, and
> 
> Attaching a.txt
> 
> > please accompany that file with instruction for reproducing starting
> > from "emacs -Q".
> 
> 0. emacs -Q
> 1. open a.txt
> 2. (toggle-truncate-lines)
> 3. goto the end of the last line
> 4. make the frame narrow so that "あ"s are not shown
> 5. click on the last word "dog"
> 
> But, this assumes that Japanese and ASCII chars have different heights.
> 
> > If it is important to click the mouse at specific
> 
> You can click anywhere on the last line.

Thanks, I see the problem and will look into it.

Re: Location of mouse click

[Eli Zaretskii]

> Date: Wed, 13 Apr 2022 18:58:14 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-devel@gnu.org
> 
> > 0. emacs -Q
> > 1. open a.txt
> > 2. (toggle-truncate-lines)
> > 3. goto the end of the last line
> > 4. make the frame narrow so that "あ"s are not shown
> > 5. click on the last word "dog"
> > 
> > But, this assumes that Japanese and ASCII chars have different heights.
> > 
> > > If it is important to click the mouse at specific
> > 
> > You can click anywhere on the last line.
> 
> Thanks, I see the problem and will look into it.

I hope I've now fixed this on the master branch.

Re: Location of mouse click

[Yasushi SHOJI]

On Fri, Apr 15, 2022 at 2:43 AM Eli Zaretskii <eliz@gnu.org> wrote:
> > > 0. emacs -Q
> > > 1. open a.txt
> > > 2. (toggle-truncate-lines)
> > > 3. goto the end of the last line
> > > 4. make the frame narrow so that "あ"s are not shown
> > > 5. click on the last word "dog"
> > >
> > > But, this assumes that Japanese and ASCII chars have different heights.
> > >
> > > > If it is important to click the mouse at specific
> > >
> > > You can click anywhere on the last line.
> >
> > Thanks, I see the problem and will look into it.
>
> I hope I've now fixed this on the master branch.

Thank you, Eli.  The change, 67e7870a, works perfectly.

Would you mind backporting to 28? I've tested the change
by cherry picking on 28, and it works without a glitch.

Thanks,
-- 
             yashi

Re: Location of mouse click

[Eli Zaretskii]

> From: Yasushi SHOJI <yasushi.shoji@gmail.com>
> Date: Fri, 15 Apr 2022 14:11:23 +0900
> Cc: emacs-devel@gnu.org
> 
> On Fri, Apr 15, 2022 at 2:43 AM Eli Zaretskii <eliz@gnu.org> wrote:
> > > > 0. emacs -Q
> > > > 1. open a.txt
> > > > 2. (toggle-truncate-lines)
> > > > 3. goto the end of the last line
> > > > 4. make the frame narrow so that "あ"s are not shown
> > > > 5. click on the last word "dog"
> > > >
> > > > But, this assumes that Japanese and ASCII chars have different heights.
> > > >
> > > > > If it is important to click the mouse at specific
> > > >
> > > > You can click anywhere on the last line.
> > >
> > > Thanks, I see the problem and will look into it.
> >
> > I hope I've now fixed this on the master branch.
> 
> Thank you, Eli.  The change, 67e7870a, works perfectly.

Thanks for testing, and for a clear test case.

> Would you mind backporting to 28? I've tested the change
> by cherry picking on 28, and it works without a glitch.

You are free to use it in your Emacs 28, but I don't consider it safe
enough to have this change in Emacs 28.2 for the general consumption.
As simple and innocent as the change may look, it was done in a very
low-level subroutine, which is used all over the place in display
code.  I cannot predict what issues it could cause elsewhere, which
would require followup changes to unbreak.  OTOH, the problem you
described exists since Emacs 25.1, so it cannot be too annoying, or we
would have heard about it sooner; thus, fixing it is not an urgent
enough need, and doesn't outweigh the potential dangers of
destabilizing the release branch.  Sorry.

Re: Location of mouse click

[Yasushi SHOJI]

On Fri, Apr 15, 2022 at 3:23 PM Eli Zaretskii <eliz@gnu.org> wrote:
> > Would you mind backporting to 28? I've tested the change
> > by cherry picking on 28, and it works without a glitch.
>
> You are free to use it in your Emacs 28, but I don't consider it safe
> enough to have this change in Emacs 28.2 for the general consumption.
> As simple and innocent as the change may look, it was done in a very
> low-level subroutine, which is used all over the place in display
> code.  I cannot predict what issues it could cause elsewhere, which
> would require followup changes to unbreak.  OTOH, the problem you
> described exists since Emacs 25.1, so it cannot be too annoying, or we
> would have heard about it sooner; thus, fixing it is not an urgent
> enough need, and doesn't outweigh the potential dangers of
> destabilizing the release branch.  Sorry.

Fair enough.

Thanks for your prompt fix, Eli.

Best regards,
-- 
            yashi