bug#9324: 24.0.50; Movement past end of screen causes weird jump

[Ivan Andrus <darthandrus@gmail.com>]

On Aug 19, 2011, at 9:06 PM, Eli Zaretskii wrote:
>> From: Ivan Andrus <darthandrus@gmail.com>
>> Date: Fri, 19 Aug 2011 20:38:09 +0200
>> Cc: antoine.levitt@gmail.com,
>> 9324@debbugs.gnu.org
>> 
>> Okay, I think I can reliably reproduce it now.  Sorry it took so long to get it right.  Anyway it seems to require both highlight-parentheses-mode and hl-sexp-mode as well as a custom mode-line face.  I invoke it with:
>> 
>> emacs -Q -l ~/.emacs.d/elpa/highlight-parentheses-1.0.1/highlight-parentheses.el -l ~/.emacs.d/local/hl-sexp.el ~/vcs/emacs/bug-example-3.el
> 
> Sorry to disappoint you, but I cannot reproduce this even with this
> precise recipe.  I downloaded the latest hl-sexp.el and
> highlight-parentheses.el, and used them exactly as shown, albeit the
> leading directories were different, and I still cannot see the
> problem.  Emacs behaves as expected.

Aaargh!  After running the progn, do you see all of line starting with 208, or just the continuation of it.  How about increasing the :line-width?  If I increase it to 8 I stop seeing the buggy behavior, so maybe at 4 you can see it?  

I was able to step through redisplay_window and watch where point changes from 
BUF PT: 2418 of 1..2419 GAP: 2419 SZ=2000
to
BUF PT: 1224 of 1..2419 GAP: 2419 SZ=2000

Here's what I did.  I have done it a few times now, so if you need me to look at other data I can.

  cd ~/vcs/emacs/local/src/
  gdb $EMACS
  break xdisp.c:14951
  break xdisp.c:14939
  break xdisp.c:14946
  run -Q -l ~/.emacs.d/elpa/highlight-parentheses-1.0.1/highlight-parentheses.el -l ~/.emacs.d/local/hl-sexp.el ~/vcs/emacs/bug-example-3.el

interact with emacs evaluating the font and progn blocks continuing as necessary.  Evaluate the forward-sexp and then start stepping through with emacs:
  
Breakpoint 3, redisplay_window (window=4307577381, just_this_one_p=0) at xdisp.c:14951
14951	      if (new_vpos >= 0)
(gdb) p new_vpos
$1 = 336
(gdb) list
14946		  new_vpos = window_box_height (w);
14947		}
14948	
14949	      /* If we need to move point for either of the above reasons,
14950		 now actually do it.  */
14951	      if (new_vpos >= 0)
14952		{
14953		  struct glyph_row *row;
14954	
14955		  row = MATRIX_FIRST_TEXT_ROW (w->desired_matrix);
(gdb) 
14956		  while (MATRIX_ROW_BOTTOM_Y (row) < new_vpos)
14957		    ++row;
14958	
14959		  TEMP_SET_PT_BOTH (MATRIX_ROW_START_CHARPOS (row),
14960				    MATRIX_ROW_START_BYTEPOS (row));
14961	
14962		  if (w != XWINDOW (selected_window))
14963		    set_marker_both (w->pointm, Qnil, PT, PT_BYTE);
14964		  else if (current_buffer == old)
14965		    SET_TEXT_POS (lpoint, PT, PT_BYTE);
(gdb) n
14955		  row = MATRIX_FIRST_TEXT_ROW (w->desired_matrix);
(gdb) n
14956		  while (MATRIX_ROW_BOTTOM_Y (row) < new_vpos)
(gdb) prow
y=0 x=0 pwid=112 a+d=12+3=15 phys=12+3=15 vis=15  L=0 T=16 R=0
start=90 end=106 ENA DISP
(gdb) n 40
14956		  while (MATRIX_ROW_BOTTOM_Y (row) < new_vpos)
(gdb) n
14957		    ++row;
(gdb) 
14956		  while (MATRIX_ROW_BOTTOM_Y (row) < new_vpos)
(gdb) 
14957		    ++row;
(gdb) 
14956		  while (MATRIX_ROW_BOTTOM_Y (row) < new_vpos)
(gdb) 
14959		  TEMP_SET_PT_BOTH (MATRIX_ROW_START_CHARPOS (row),
(gdb) prow
y=330 x=0 pwid=105 a+d=12+3=15 phys=12+3=15 vis=15  L=0 T=15 R=0
start=1224 end=1239 ENA DISP
(gdb) ppt
BUF PT: 2418 of 1..2419 GAP: 2419 SZ=2000
(gdb) n
14962		  if (w != XWINDOW (selected_window))
(gdb) prow
y=330 x=0 pwid=105 a+d=12+3=15 phys=12+3=15 vis=15  L=0 T=15 R=0
start=1224 end=1239 ENA DISP
(gdb) ppt
BUF PT: 1224 of 1..2419 GAP: 2419 SZ=2000


To be honest I'm not sure how new_vpos was set since the other breakpoints are not triggered after having evaluated forward-sexp.  The breakpoint on "new_vpos = window_box_height (w) / 2;" is hit, but after evaluating the progn, so I'm confused.  Of course I have no idea if I'm even asking the right questions, but maybe you can tell me what else I need to look at.

-Ivan