bug#12364: 24.2.50; wrong overhang display for gstring compositions

bug#12364: 24.2.50; wrong overhang display for gstring compositions

[YAMAMOTO Mitsuharu]

Steps to Reproduce:

1. $ emacs -Q &
2. evaluate (set-fontset-font t 'arabic (font-spec :family "arial" :size 30))
3. C-u C-\ arabic RET
4. u S u
   At this moment, the two grapheme clusters look connected.
5. C-b C-p
   Now the two grapheme clusters look unconnected.

See also the "u-S-u" example in the first screenshot in
http://lists.gnu.org/archive/html/bug-gnu-emacs/2012-09/msg00178.html .

I think at least the two occurrences of the following line in xdisp.c
are wrong.

      int first = (s->first_glyph - glyphs) + (s->cmp ? 1 : s->nchars);

The value of s->cmp seems to be NULL for the "gstring" cases.

					YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

In GNU Emacs 24.2.50.1 (x86_64-apple-darwin12.1.0, X toolkit, Xaw3d scroll bars)
 of 2012-09-06 on YAMAMOTO-no-iMac.local
Bzr revision: 109895 yamaoka@jpl.org-20120905224543-v2x5qiqin2b83lnv
Windowing system distributor `The X.Org Foundation', version 11.0.11204000
Configured using:
 `configure '--enable-checking' '--with-jpeg=no' '--with-gif=no'
 '--with-tiff=no' 'CFLAGS=-g3''

Important settings:
  value of $LANG: ja_JP.UTF-8
  locale-coding-system: utf-8-unix
  default enable-multibyte-characters: t

Major mode: Lisp Interaction

Minor modes in effect:
  tooltip-mode: t
  mouse-wheel-mode: t
  tool-bar-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

bug#12364: 24.2.50; wrong overhang display for gstring compositions

[Eli Zaretskii]

> Date: Thu, 06 Sep 2012 11:37:26 +0900
> From: YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>
> 
> Steps to Reproduce:
> 
> 1. $ emacs -Q &
> 2. evaluate (set-fontset-font t 'arabic (font-spec :family "arial" :size 30))
> 3. C-u C-\ arabic RET
> 4. u S u
>    At this moment, the two grapheme clusters look connected.
> 5. C-b C-p
>    Now the two grapheme clusters look unconnected.

Not reproducible on Windows XP.

Are you using the latest libm17n-flt and libotf libraries, and the
latest fonts?

> I think at least the two occurrences of the following line in xdisp.c
> are wrong.
> 
>       int first = (s->first_glyph - glyphs) + (s->cmp ? 1 : s->nchars);
> 
> The value of s->cmp seems to be NULL for the "gstring" cases.

These two lines are not executed here, when the above recipe is used.
Maybe that's why I don't see the problem.

bug#12364: 24.2.50; wrong overhang display for gstring compositions

[YAMAMOTO Mitsuharu]

[-- Attachment #1: Type: text/plain, Size: 2759 bytes --]

>>>>> On Thu, 06 Sep 2012 11:37:26 +0900, YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> said:

> Steps to Reproduce:
> 1. $ emacs -Q &
> 2. evaluate (set-fontset-font t 'arabic (font-spec :family "arial" :size 30))
> 3. C-u C-\ arabic RET
> 4. u S u
>    At this moment, the two grapheme clusters look connected.
> 5. C-b C-p
>    Now the two grapheme clusters look unconnected.

> See also the "u-S-u" example in the first screenshot in
> http://lists.gnu.org/archive/html/bug-gnu-emacs/2012-09/msg00178.html .

> I think at least the two occurrences of the following line in xdisp.c
> are wrong.

>       int first = (s->first_glyph - glyphs) + (s->cmp ? 1 : s->nchars);

> The value of s->cmp seems to be NULL for the "gstring" cases.

I tried making a fix.  The change for xterm.c is so the box cursor can
invert the left overhang of the succeeding glyph.  The attached
screenshots show the difference between with (first) and without
(second) the patch with respect to the following steps.

1. emacs -Q -D &
2. (set-frame-font "Times-32") C-j
3. C-u y
4. C-a C-SPC C-e M-o b
5. C-a C-SPC C-e M-o i
6. C-b C-b

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

=== modified file 'src/xdisp.c'
*** src/xdisp.c	2012-09-05 03:37:32 +0000
--- src/xdisp.c	2012-09-07 01:21:28 +0000
***************
*** 23050,23056 ****
      {
        int x = 0, i;
        struct glyph *glyphs = s->row->glyphs[s->area];
!       int first = (s->first_glyph - glyphs) + (s->cmp ? 1 : s->nchars);
        int end = s->row->used[s->area];
  
        for (i = first; i < end && s->right_overhang > x; ++i)
--- 23050,23057 ----
      {
        int x = 0, i;
        struct glyph *glyphs = s->row->glyphs[s->area];
!       int first = (s->first_glyph - glyphs
! 		   + (s->first_glyph->type == COMPOSITE_GLYPH ? 1 : s->nchars));
        int end = s->row->used[s->area];
  
        for (i = first; i < end && s->right_overhang > x; ++i)
***************
*** 23073,23079 ****
    int i, k, x;
    int end = s->row->used[s->area];
    struct glyph *glyphs = s->row->glyphs[s->area];
!   int first = (s->first_glyph - glyphs) + (s->cmp ? 1 : s->nchars);
  
    k = -1;
    x = 0;
--- 23074,23081 ----
    int i, k, x;
    int end = s->row->used[s->area];
    struct glyph *glyphs = s->row->glyphs[s->area];
!   int first = (s->first_glyph - glyphs
! 	       + (s->first_glyph->type == COMPOSITE_GLYPH ? 1 : s->nchars));
  
    k = -1;
    x = 0;

=== modified file 'src/xterm.c'
*** src/xterm.c	2012-09-07 01:27:44 +0000
--- src/xterm.c	2012-09-07 03:41:30 +0000
***************
*** 2975,2980 ****
--- 2975,2981 ----
  		XSetClipMask (next->display, next->gc, None);
  		next->hl = save;
  		next->num_clips = 0;
+ 		next->clip_head = s->next;
  	      }
  	}
      }


[-- Attachment #2: after.png --]
[-- Type: image/png, Size: 4442 bytes --]

[-- Attachment #3: before.png --]
[-- Type: image/png, Size: 4412 bytes --]

bug#12364: 24.2.50; wrong overhang display for gstring compositions

[YAMAMOTO Mitsuharu]

>>>>> On Fri, 07 Sep 2012 17:47:23 +0900, YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> said:

> I tried making a fix.  The change for xterm.c is so the box cursor can
> invert the left overhang of the succeeding glyph.

I've just installed the patch.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp