RE: [External] : Re: Concern about new binding.

[Drew Adams <drew.adams@oracle.com>]

> >> ... but apparently you prefer to continue to use the
> >> few remaining keys that are not bound by default?
> >> Isn't that contradictory?
> >
> > How so?  The few remaining keys are more than a single key.
> 
> ... and a complete keymap is more than a single key.

No, sorry; I still don't understand.  You can bind
a keymap to a key.  If there are _several_ keys that
you can bind keymaps to, then that offers more
possibilities than if there is only _one_ key that
you can bind a keymap to.

> I can't understand why you do not agree that having
> _all_ (say) C-o LETTER slots at the disposal of
> third-party libraries is not a reasonable solution
> that would contribute to peace and stability, and
> that you prefer to fight to keep the last five or
> six C-x LETTER slots free.

`C-o' is currently bound by default.  But I already
made clear that I'm for freeing it up, one way or
another.  I _do_ agree that having all `C-o LETTER'
keys available for binding would be helpful.

If _only_ `C-o' were available for use as a prefix key
then that would be far less than all of the currently
unbound keys, which can each be used as a prefix key.

And the currently unbound keys are not limited to
`C-x LETTER' or even `C-x <whatever>' keys.

I proposed a moratorium on Emacs binding any new keys
by default.  I didn't limit that to `C-x LETTER'.

And yes, we could go further, and consider adding some
currently globally bound keys, such as `C-o', to the
"free" list.  I wouldn't be against that, at all, even
if I only argued for Emacs to keep its mitts off keys
keys that don't yet have global default bindings.

My proposal is a good start.  But I sure wouldn't mind
having us go further.  Emacs has too many keys bound
globally by default, I think.  That latter battle
isn't one I'd argue for now, but yeah, I'm in favor
of Emacs having even fewer global key bindings, not
just in favor of it putting an end to the steal.
(Ooooh, apologies for that horrible turn of phrase.)

Something needs to be done, IMO.  Other proposals are
welcome, to help us manage key-binding possibilities
and their inherent limitations.  Early on, there was
a vast undeveloped and unexplored frontier - uncharted
territory.  Now, Land's End is in plain sight.

Maybe Emacs should bind global keys only using some
hydra-like functionality.  (Dunno.)

But whatever solution we might find, it's important
that the help system really support it.  Today, help
on keys is good - even static help, i.e., without
something like `which-key' or Icicles key completion.
Today, you can do `C-x 4 C-h' and get help on all of
those keys.  Likewise, `describe-keymap' is a great
help.  We need to ensure that we keep providing great
key help.

Dunno whether a hydra approach - or even some of the
currently discussed transient key approaches -
provide such good help.  I suspect not, but I don't
know.  It's one thing to provide on-the-fly help
with some kind of completion when you start to use
a key.  It's something else to be able to get help
about whole sets of keys.