bug#13277: 24.2.91; -nw display bug with overlay-arrow

bug#13277: 24.2.91; -nw display bug with overlay-arrow

[Andreas Schwab]

This is a regression from emacs 23, already present in 24.1:

$ cat bug.el
(set (make-local-variable 'overlay-arrow-string) "=>")
(set (make-local-variable 'overlay-arrow-position) (point-marker))
;;(setq truncate-lines t)
(insert-char ?- (window-width))
(goto-char (point-min))
$ emacs -nw -l ./bug.el

The cursor is placed at the right margin.  Either moving forward two
characters or deleting one character (so that the line fits in the
window) puts the cursor after the arrow, where it belongs.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#13277: 24.2.91; -nw display bug with overlay-arrow

[Eli Zaretskii]

> From: Andreas Schwab <schwab@linux-m68k.org>
> Date: Tue, 25 Dec 2012 19:13:44 +0100
> 
> This is a regression from emacs 23, already present in 24.1:
> 
> $ cat bug.el
> (set (make-local-variable 'overlay-arrow-string) "=>")
> (set (make-local-variable 'overlay-arrow-position) (point-marker))
> ;;(setq truncate-lines t)
> (insert-char ?- (window-width))
> (goto-char (point-min))
> $ emacs -nw -l ./bug.el
> 
> The cursor is placed at the right margin.  Either moving forward two
> characters or deleting one character (so that the line fits in the
> window) puts the cursor after the arrow, where it belongs.

It's a consequence of the bidirectional display, under which we can no
longer assume that, if point is at position P, and that position is
not visible, the cursor should be put on the first visible glyph whose
position is >= P.  (In this case, the first 2 characters are obscured
by the overlay arrow, and don't appear on the line.)

I will look into fixing this soon.  Thanks for reporting this.

bug#13277: 24.2.91; -nw display bug with overlay-arrow

[Eli Zaretskii]

> Date: Tue, 25 Dec 2012 21:03:22 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 13277@debbugs.gnu.org
> 
> > From: Andreas Schwab <schwab@linux-m68k.org>
> > Date: Tue, 25 Dec 2012 19:13:44 +0100
> > 
> > This is a regression from emacs 23, already present in 24.1:
> > 
> > $ cat bug.el
> > (set (make-local-variable 'overlay-arrow-string) "=>")
> > (set (make-local-variable 'overlay-arrow-position) (point-marker))
> > ;;(setq truncate-lines t)
> > (insert-char ?- (window-width))
> > (goto-char (point-min))
> > $ emacs -nw -l ./bug.el
> > 
> > The cursor is placed at the right margin.  Either moving forward two
> > characters or deleting one character (so that the line fits in the
> > window) puts the cursor after the arrow, where it belongs.
> 
> It's a consequence of the bidirectional display, under which we can no
> longer assume that, if point is at position P, and that position is
> not visible, the cursor should be put on the first visible glyph whose
> position is >= P.  (In this case, the first 2 characters are obscured
> by the overlay arrow, and don't appear on the line.)

Should be fixed in revision 111082 on the emacs-24 branch.  (The fix
is a simple one-liner, and is so obviously correct that it can only
cause problems if it exposes another bug, so I think it's appropriate
for the branch.)