Re: cursor-in-non-selected-windows as buffer-local

[md5i]

address@hidden (Gerd Moellmann) writes:

> address@hidden writes:
>
>> I hadn't, but I don't think I can easily use this to do what I want.
>> 
>> For example: I like the default setting of
>> cursor-in-non-selected-windows being t.  On the other hand, I don't
>> like seeing an empty-box cursor in my Gnus article buffers.  When I
>> am actually *in* the article buffer, however, I want my normal box
>> cursor.
>
> I think one could easily add a `cursor-type' of, say,
> `selected-window' with that effect.  It seems easier to do than to
> make `cursor-in-non-selected-windows' buffer-local and extend it to
> contain a list of buffers or something.

That would be a possible solution to the problem, yes.  Although I
don't know that it is the proper thing to do semantically.  If you
added this as a cursor-type, then you couldn't have my desired
behavior with the empty-boxes along with a bar cursor.  Unless you
want to extend cursor-type to include things like 
(selected-window . (bar . WIDTH)).

On the other hand, your comment on a list of buffers makes me think
that you didn't understand my proposition:

Proposition: Within the function of x_display_and_set_cursor,
instead of using the value of cursor_in_non_selected_windows, we use
its buffer-local value (if any) in the buffer of the window for which
we are drawing the cursor.  So, in order to get the effect I am
looking for, one has only to do the following when creating the
buffer:

(make-variable-buffer-local 'cursor-in-non-selected-windows)
(setq cursor-in-non-selected-windows nil)

This proposition loses if we like the current behavior when making
cursor-in-non-selected-windows buffer-local, but I don't know that
anyone has even considered doing that, other than myself.

-- 
Michael Welsh Duggan
(address@hidden)