bug#12170: save-excursion fails boundary case with recenter

[martin rudalics]

> (1) By "window height", I meant the number of lines displayed for the actual
> buffer, not counting the mode line or minibuffer.  It turns out this is one
> less than the value returned by `window-height'.
>
> (2) The value of `point' changes (the cursor "hops", in other words) when
> the window is redisplayed, or when control returns to the user.
>
> Here is code that will produce the bug (with emacs -Q):
>
> (progn
>    (defun f (n)
>      (save-excursion (forward-line (- n)) (recenter 0)))
>    (let ((buffer (switch-to-buffer "foo"))
>          (height (1- (window-height (get-buffer-window "foo")))))
>      (insert-char 10 (* height 2))
>      (let ((pt (point)))
>        (f height)
>        (redisplay)
>        (message "height %s old %s new %s" height pt (point)))))
>
> If you leave out the `redisplay' call, on the other hand, old = new in the
> message -- but after control returns to the user, point has still been moved
> in the buffer.

I see it now, thanks.  Surprisingly it doesn't show up with code like

(progn
  (defmacro save-this-window-excursion (&rest body)
    "..."
    (let ((start (make-symbol "start"))
          (point (make-symbol "point")))
      `(let ((,start (copy-marker (window-start)))
             (,point (copy-marker (window-point))))
         (save-selected-window
           (progn ,@body))
         (set-window-start (selected-window) ,start t)
         (set-window-point (selected-window) ,point))))

  (defun f (n)
     (save-this-window-excursion (forward-line (- n)) (recenter 0)))

   (let ((buffer (switch-to-buffer "foo"))
         (height (1- (window-height (get-buffer-window "foo")))))
     (insert-char 10 (* height 2))
     (let ((pt (point)))
       (f height)
       (redisplay)
       (message "height %s old %s new %s" height pt (point)))))

so I'd suspect the culprit somewhere in redisplay_window's code

  if (w->optional_new_start

w->optional_new_start is 1 from `recenter'

      && CHARPOS (startp) >= BEGV
      && CHARPOS (startp) <= ZV)
    {
      w->optional_new_start = 0;
      start_display (&it, w, startp);
      move_it_to (&it, PT, 0, it.last_visible_y, -1,
                  MOVE_TO_POS | MOVE_TO_X | MOVE_TO_Y);
      if (IT_CHARPOS (it) == PT)
        w->force_start = 1;
      /* IT may overshoot PT if text at PT is invisible.  */
      else if (IT_CHARPOS (it) > PT && CHARPOS (startp) <= PT)
        w->force_start = 1;

w->force_start 1 will cause redisplay to honor the start position set up
by `recenter' despite of save_excursion_restore's Fgoto_char.

    }

But I don't have the slightest idea how calling

         (set-window-start (selected-window) ,start t)

would remedy this.  Eli will soon teach us a lesson here.

martin