Package: emacs;
Reported by: Keith David Bershatsky <esq <at> lawlist.com>
Date: Tue, 23 Jan 2018 07:33:02 UTC
Severity: minor
Tags: wontfix
Done: Stefan Kangas <stefankangas <at> gmail.com>
Bug is archived. No further changes may be made.
View this message in rfc822 format
From: Eli Zaretskii <eliz <at> gnu.org> To: Keith David Bershatsky <esq <at> lawlist.com> Cc: 30226 <at> debbugs.gnu.org Subject: bug#30226: Fixing it->pixel_width / it->current_x when tabs and line numbers. Date: Sat, 27 Jan 2018 16:56:05 +0200
> Date: Mon, 22 Jan 2018 23:52:20 -0800
> From: Keith David Bershatsky <esq <at> lawlist.com>
>
> Here is a revised patch that removes an erroneous section that was experimental.
Thanks, but I'm unable to reproduce the original problem, so I cannot
yet discuss the reasons, let alone the solution.
Here's what I did in Emacs built from the current emacs-26 branch:
1) Inserted the following snippet into *scratch*:
(setq display-line-numbers t)
(setq buffer-display-table (make-display-table))
(aset buffer-display-table
?\t
(vector (make-glyph-code ?\u00BB 'font-lock-warning-face)
(make-glyph-code ?\t 'highlight)))
(setq tab-width 8)
2) Evaluated it ("M-x eval-region RET").
3) Went to the end of the buffer, inserted a few empty lines, then
went up 2 lines from EOB, and typed "C-q TAB Hello-world"
4) Typed "M-x dump-glyph-row RET"
The result is this data shown on stderr:
Row Start End Used oE><\CTZFesm X Y W H V A P
==============================================================================
11 409 422 18 010000100000 0 176 192 16 16 12 12
-1 -1 0
-1 -1
-1 -1
Glyph# Type Pos O W Code C Face LR
0 C -1 0 8 0x000020 20 00
1 C -1 0 8 0x000031 1 20 00
2 C -1 0 8 0x000032 2 20 00
3 C -1 0 8 0x000020 20 00
4 C 409 B 8 0x0000bb . 23 00
5 S 409 B 56 0x000000 22 00 <<<<<<<<<<<<<<<<<<
6 C 410 B 8 0x000048 H 0 00
7 C 411 B 8 0x000065 e 0 00
8 C 412 B 8 0x00006c l 0 00
9 C 413 B 8 0x00006c l 0 00
10 C 414 B 8 0x00006f o 0 00
11 C 415 B 8 0x00002d - 0 00
12 C 416 B 8 0x000077 w 0 00
13 C 417 B 8 0x00006f o 0 00
14 C 418 B 8 0x000072 r 0 00
15 C 419 B 8 0x00006c l 0 00
16 C 420 B 8 0x000064 d 0 00
17 C 0 0 8 0x000020 0 00
This is as expected: the width of the default font is 8, and the size
of the stretch glyph #5 is 8x8 = 56, as expected.
Next, I typed "M-: (scroll-left 1) RET" and again invoked
dump-glyph-row. The result:
Row Start End Used oE><\CTZFesm X Y W H V A P
==============================================================================
11 409 422 17 011000100000 0 176 176 16 16 12 12
-1 -1 0
-1 -1
-1 -1
Glyph# Type Pos O W Code C Face LR
0 C -1 0 8 0x000020 20 00
1 C -1 0 8 0x000031 1 20 00
2 C -1 0 8 0x000032 2 20 00
3 C -1 0 8 0x000020 20 00
4 S 409 B 48 0x000000 22 00 <<<<<<<<<<<<<<
5 C 410 B 8 0x000048 H 0 00
6 C 411 B 8 0x000065 e 0 00
7 C 412 B 8 0x00006c l 0 00
8 C 413 B 8 0x00006c l 0 00
9 C 414 B 8 0x00006f o 0 00
10 C 415 B 8 0x00002d - 0 00
11 C 416 B 8 0x000077 w 0 00
12 C 417 B 8 0x00006f o 0 00
13 C 418 B 8 0x000072 r 0 00
14 C 419 B 8 0x00006c l 0 00
15 C 420 B 8 0x000064 d 0 00
16 C 0 0 8 0x000020 0 00
Note that the glyph corresponding to the u+00BB character now
disappeared from display, and therefore the stretch glyph is #4, its
width is 48 -- again, as expected.
I then typed "M-: (scroll-left 1) RET" once more, followed by
dump-glyph-row, and the results are:
Row Start End Used oE><\CTZFesm X Y W H V A P
==============================================================================
11 409 422 17 011000100000 0 176 168 16 16 12 12
-1 -1 0
-1 -1
-1 -1
Glyph# Type Pos O W Code C Face LR
0 C -1 0 8 0x000020 20 00
1 C -1 0 8 0x000031 1 20 00
2 C -1 0 8 0x000032 2 20 00
3 C -1 0 8 0x000020 20 00
4 S 409 B 40 0x000000 22 00 <<<<<<<<<<<<<
5 C 410 B 8 0x000048 H 0 00
6 C 411 B 8 0x000065 e 0 00
7 C 412 B 8 0x00006c l 0 00
8 C 413 B 8 0x00006c l 0 00
9 C 414 B 8 0x00006f o 0 00
10 C 415 B 8 0x00002d - 0 00
11 C 416 B 8 0x000077 w 0 00
12 C 417 B 8 0x00006f o 0 00
13 C 418 B 8 0x000072 r 0 00
14 C 419 B 8 0x00006c l 0 00
15 C 420 B 8 0x000064 d 0 00
16 C 0 0 8 0x000020 0 00
Now the stretch glyph's width is 40 pixels, again as expected.
So the width of the stretch glyph is computed correctly. And your
screenshots also show similar results, where the width of the stretch
glyph is decremented for each call to scroll-left.
Your problem description talks about it->pixel_width, not about
glyph->pixel_width, but the latter should be equal to the former,
because x_produce_glyphs does this when it finishes computing the
width of the stretch glyph:
if (it->glyph_row)
{
append_stretch_glyph (it, it->object, it->pixel_width,
it->ascent + it->descent, it->ascent);
and the function append_stretch_glyph then assigns the value of its
3rd argument to glyph->pixel_width:
static void
append_stretch_glyph (struct it *it, Lisp_Object object,
int width, int height, int ascent)
{
...
glyph->pixel_width = clip_to_bounds (-1, width, SHRT_MAX);
So I'm not sure why you are saying there's a bug in the display code
regarding it->pixel_width of the stretch glyphs that represent TAB
characters.
You say:
> In a nutshell, it->pixel_width and it->current_x are both incorrect in that situation. Because the X is wrong, all subsequent references to it->current_x on the same line are also wrong.
I don't understand why you are saying that it->current_x is wrong.
What is that assertion based on? Maybe there's some misunderstanding
of what current_x is and relative to what position does it measure the
X coordinate. Can you tell what you expected current_x to be in some
specific situation (i.e., specific horizontal scroll of the display
in the above scenario), and what you really found?
GNU bug tracking system
Copyright (C) 1999 Darren O. Benham,
1997,2003 nCipher Corporation Ltd,
1994-97 Ian Jackson.