Re: more problems with line-move

[Richard Stallman]

    If point was at position A and is going to position B and both positions map
    to the same screen position, then my code moves point to B+1 (or B-1
    depending on the direction, of course).  That's not always correct, but at
    that level (i.e. without knowing what the command intended to do) it's
    difficult to do any better.

That avoids the "point doesn't appear to move" inconsistency, but
other inconsistent behavior will still occur.  On two occasions where
point appears to be at the same place, it will really be at A and at
B.  C-k will kill different text.

I designed my invisible-intangible patch to prevent such inconsistent
behavior.  The idea is that position B is part of the range of
off-limits positions, even though the character between B and B+1 is
neither invisible or intangible.  Therefore, the inconsistency of
A or B cannot happen.

    But yes, we can sometimes reach point A and sometimes point B (both ends of
    a non-sticky invisible extent of text) depending on how the user gets to
    this text.  For sticky invisible text, this should not happen.

How does stickiness affect it?  Perhaps telling people to use stickiness
in certain ways is a solution.  Or perhaps this code should be changed
in regard to how it depends on stickiness.