Line Continuation Bugs & Patch

Line Continuation Bugs & Patch

[Ein Terakawa]

Emacs-21 currently has a number of bugs concerning line continuation.
That is mainly in the code which handle !it->truncate_lines_p.
I checked Emacs-21.1 , Emacs-21.2 and one in the cvs tree (21.2.50).

I admit this bug report does not conform to the guideline.
But It was much fun to fix bugs than to report it and just wait.
So I created a patch. Although it may not be well prepared to be
inspected by Emacs core developers, it fixes bugs. And Emacs works
as fine as not fixed one dose.

Here, I just wanted to announce that there are bugs and I'm working on it.
If it's needed, I'm willing to discuss the problems on emacs-devel.
Boring bug description should be done first, though.


This is the patch. 
http://applause.elfmimi.jp/emacs-lisp/emacs-21.1-proper-continuation-20020518.diff.gz
And some documentation in my mother tongue (Japanese).
http://applause.elfmimi.jp/emacs-lisp/emacs-21.1-proper-continuation-20020518-ja.txt

Structs, macros and functions get affected by:
 struct glyph_row
 MATRIX_ROW_CONTINUATION_LINE_P
 MATRIX_ROW_ENDS_IN_MIDDLE_OF_CHAR_P
 compute_motion 
 buffer_posn_from_coords
 Fvertical_motion
 window_scroll_pixel_based
 start_display
 move_it_in_display_line_to
 move_it_to
 move_it_vertically_backward
 move_it_by_lines
 try_window
 try_window_id
 display_line
----
Ein Terakawa <applause@elfmimi.jp>

Re: Line Continuation Bugs & Patch

[Richard Stallman]

    Emacs-21 currently has a number of bugs concerning line continuation.
    That is mainly in the code which handle !it->truncate_lines_p.
    I checked Emacs-21.1 , Emacs-21.2 and one in the cvs tree (21.2.50).

Could you send us the patches that apply to Emacs 21.2 and to the CVS
tree (21.2.50)?  We would probably like to install the patches.

Re: Line Continuation Bugs & Patch

[Ein Terakawa]

Well, I just created patches for 21.2 and 21.2.50 .

There is one thing I'm not clear.
That is the follwing line in window_scroll_pixel_based of window.c .
      w->start_at_line_beg = Fbolp ();
I think we can't say  PT == window-start  here. Am I right?

I'll be happy if you can decide to install the patch just by looking at
the chages the patch makes, and if I don't have to do things like detailed
description. If so I'can just proceed to the farther step of line
continuation issue.
Please let me know if there is something I'm supposed to do.


Notice:

* As for 21.2.50 the patch intentionally overrides a change
  done by gerd at revision 1.150 of indent.c .

* Some messy keywords ('XXX' and 'XXXX') are contained.

* No change logs.


Here are the new patches.
http://applause.elfmimi.jp/emacs-lisp/emacs-21.1-proper-continuation-20020520.diff.gz
http://applause.elfmimi.jp/emacs-lisp/emacs-21.2-proper-continuation-20020520.diff.gz
http://applause.elfmimi.jp/emacs-lisp/emacs-21.2.50-proper-continuation-20020520.diff.gz
# Thanks to the flexibility of the patch command
# patch for 21.1 can also be applied to 21.2 .

----
Ein Terakawa <applause@elfmimi.jp>

Re: Line Continuation Bugs & Patch

[Richard Stallman]

    That is the follwing line in window_scroll_pixel_based of window.c .
	  w->start_at_line_beg = Fbolp ();
    I think we can't say  PT == window-start  here. Am I right?

You are right.  So it is a bug to use Fbolp there.
Does this fix it?

*** window.c.~1.414.~	Sat Apr 27 15:54:02 2002
--- window.c	Sun May 26 14:30:50 2002
***************
*** 4162,4171 ****
  
    if (! vscrolled)
      {
        /* Set the window start, and set up the window for redisplay.  */
!       set_marker_restricted (w->start, make_number (IT_CHARPOS (it)),
  			     w->buffer);
!       w->start_at_line_beg = Fbolp ();
        w->update_mode_line = Qt;
        XSETFASTINT (w->last_modified, 0);
        XSETFASTINT (w->last_overlay_modified, 0);
--- 4162,4175 ----
  
    if (! vscrolled)
      {
+       int pos = IT_CHARPOS (it);
+       int bytepos;
        /* Set the window start, and set up the window for redisplay.  */
!       set_marker_restricted (w->start, make_number (pos),
  			     w->buffer);
!       bytepos = XMARKER (w->start)->bytepos;
!       w->start_at_line_beg = ((pos == BEGV || FETCH_BYTE (bytepos - 1) == '\n')
! 			      ? Qt : Qnil);
        w->update_mode_line = Qt;
        XSETFASTINT (w->last_modified, 0);
        XSETFASTINT (w->last_overlay_modified, 0);

Re: Line Continuation Bugs & Patch

[Ein Terakawa]

> Does this fix it?
Yes, it is mostly the same change that I included in the patch.
Besides, it didn't seem to be causing any noticeable bug for a user.
At this point of time I mean.

----
Ein Terakawa <applause@elfmimi.jp>