Overlays at the end of the buffer does not reach it

Overlays at the end of the buffer does not reach it

[Lennart Borgman (gmail)]

If you put an overlay at the end of the buffer and then put point at the 
  end of the buffer then point is outside the overlay.

That prevents the use of overlay keymaps at the end of the buffer. I 
therefor suggest that in this case the point should be regarded to be 
within the overlay.

Or, alternatively that move-overlay used with something like

   (widen)
   (move-overlay ovl beg (1+ (point-max)))

should make the overlay behave the way I suggest.

However when inserting text at the end of the buffer I guess that text 
still should be regarded as placed after the overlay. Otherwise it would 
be very inconvenient to add text after the overlay.

Re: Overlays at the end of the buffer does not reach it

[Stefan Monnier]

> If you put an overlay at the end of the buffer and then put point at the end
> of the buffer then point is outside the overlay.

Doesn't it depend on the insertion type of the end of the overlay?
I.e. (make-overlay (point) (point-max) nil nil t) should work.
I'm not claiming it does, but if it doesn't, I think it's a bug.


        Stefan

Re: Overlays at the end of the buffer does not reach it

[Lennart Borgman (gmail)]

Stefan Monnier wrote:
>> If you put an overlay at the end of the buffer and then put point at the end
>> of the buffer then point is outside the overlay.
> 
> Doesn't it depend on the insertion type of the end of the overlay?
> I.e. (make-overlay (point) (point-max) nil nil t) should work.
> I'm not claiming it does, but if it doesn't, I think it's a bug.

Yes, it works the way you say. Thanks.

Re: Overlays at the end of the buffer does not reach it

[Richard Stallman]

    If you put an overlay at the end of the buffer and then put point at the 
      end of the buffer then point is outside the overlay.

    That prevents the use of overlay keymaps at the end of the buffer. I 
    therefor suggest that in this case the point should be regarded to be 
    within the overlay.

This issue calls for very careful thought.  The change you're proposing
would be useful, but it would be an inconsistency.  Inconsistencies
in the meaning of Lisp data tend to bite you.  For instance, what happens
if you narrow and set ZV to the place where an overlay ends.
Does that cause the overlay to start affecting the position at the end
whereas previously it did not?

If this doesn't depend on the marker insertion type of each end of the
overlay, perhaps making it depend on that is a good solution.

Re: Overlays at the end of the buffer does not reach it

[Lennart Borgman (gmail)]

Richard Stallman wrote:
> If this doesn't depend on the marker insertion type of each end of the
> overlay, perhaps making it depend on that is a good solution.

It does depend on the marker insertion type. Stefan told this and I 
sorry for the trouble I caused here. (And of course a bit ashamed I 
missed that.)