Hi, another questionable (I think) assertion failure, this time in
adjust_point_for_property (keyboard.c):

        ...
	  /* Move away from the inside area.  */
	  if (beg < PT && end > PT)
	    {
	      SET_PT ((orig_pt == PT && (last_pt < beg || last_pt > end))
                      /* ... */
		      ? (orig_pt = -1, PT < last_pt ? end : beg)
                      /* ... */
		      : (PT < last_pt ? beg : end));
	      check_composition = check_display = 1;
	    }
===>	  xassert (PT == beg || PT == end);

I'm not sure exactly what happened here (gdb sess attached below
though), but PT == 500, and end == 499.

The context is that I was trying out `line-move-ignore-invisible' (after
seeing Luc's post), setting it to `t', and viewing some messages in Gnus
where invisible text was used to elide citations.  At some point it
died (I think I was using C-n and C-p to move past the invisible area
seeing how the above variable affected things).

What I'm wondering is: is this assertion reasonable?  SET_PT can do
non-trivial things (calling hooks etc), so assuming that point ends up
being exactly what you set it too seems a bit questionable.

Thanks,

-Miles