Re: momentary-string-display

[Kevin Rodgers <kevin.d.rodgers@gmail.com>]

Richard Stallman wrote:
> Could you please try your changed version with various commands that use
> momentary-string-display, and see if it fully works?

The only calls to momentary-string-display are in ada-prj.el,
fortran.el, pascal.el, and table.el; but I don't use any of those
libraries, so I wouldn't be able to tell whether the function is
working the same, better, or worse in those contexts.

> As for this code,
> 
>     ! 	    ;; If the message end is off screen, recenter now.
>     ! 	    (if (< (window-end nil t) insert-end)
>     ! 		(recenter (/ (window-height) 2)))
>     ! 	    ;; If that pushed message start off the screen,
>     ! 	    ;; scroll to start it at the top of the screen.
>     ! 	    (move-to-window-line 0)
>     ! 	    (if (> (point) pos)
>     ! 		(progn
>     ! 		  (goto-char pos)
>     ! 		  (recenter 0))))
> 
> you could replace it with some reasonable heuristic that
> doesn't go wrong very often.

As I meant to say, the same heuristic can be used by replacing the
reference to insert-end with (+ pos (length string)) -- as long as
the new call to overlay-put causes a redisplay that updates window-end
just as the original call to insert does.

Is there a problem with the current heuristic?  I think now that I
should have left that code in (with insert-end replaced as described).

-- 
Kevin Rodgers
Denver, Colorado, USA