bug#3540: Please reserve a ctrl-key combination for interoperability

[Drew Adams]

> Actually, that's what I do in my config.  (Is there anyone who actually
> wants C-z's default binding???)

Dunno.  But I use something very close to it (for GUI).
It has the same behavior as the default command, unless
you use a prefix arg:

(defun iconify/show-frame (&optional all-action)
  "Iconify selected frame if now shown.  Show it if now iconified.
A non-negative prefix arg iconifies all shown frames.
A negative prefix arg deiconifies all iconified frames."
  (interactive "P")
  (cond ((not all-action)
         (when rename-frame-when-iconify-flag
           (rename-non-minibuffer-frame))
         (iconify-or-deiconify-frame))
        ((natnump (prefix-numeric-value all-action))
         (iconify-everything))
        (t (deiconify-everything)))) ; <== Emacs default

But I'm not arguing to keep the default `C-z' binding.
I'm really arguing against wasting `C-z' on something
else, by default.

Someday we'll come across a really important new feature
that really deserves `C-z' (e.g. as a prefix key).  Keys
shouldn't be bound by default lightly.  Once a key is
bound by default it becomes harder to later remove or
replace its binding.

`C-z' is a wonderful key for general things, including
use as a prefix key.  (And if not a prefix key then at
least for a repeatable command.)  And it's as easy to
reach as `C-x' on most keyboards.

If ever there was a key that I don't think should be
bound by default willy nilly (aka wasted, in my view),
it's `C-z'.

> > The manual should not suggest that users bind any
> > particular keys.  It's OK for a 3rd-party library
> > to suggest key bindings.  It's not good for Emacs
> > itself to do that.
> 
> I'm not sure I agree.  I'd welcome a list of bindings like C-z or M-o
> which do nothing useful by default.  (In fact, I compiled such a list
> myself - https://urldefense.proofpoint.com/v2/url?u=http-3A__mbork.pl_2019-
> 2D03-2D18-5FFree-5FEmacs-5Fkey-
> 5Fbindings&d=DwIBAg&c=RoP1YumCXCgaWHvlZYR8PZh8Bv7qIrMUB65eapI_JnE&r=kI3P6ljGv
> 6CTHIKju0jqInF6AOwMCYRDQUmqX22rJ98&m=vtPmjCas97xufIKnzS06ZEh04AKsMp8iJj-
> 9W7kHURQ&s=zXD57trzqr01K4n0EBvwEqWYU6PCjR6gahEDtLcmms8&e=  - but I'm
> not very happy with it.)

One person's not-very-useful is another's useful.

And certainly the manual shouldn't suggest that users
bind some key that has a default binding because that
binding isn't very useful.

If Emacs really thinks some default binding isn't
very useful then it shouldn't bind it by default.

> > 3rd-party libraries are opt-in by users.  Using
> > one is like adding its feature/code to your init
> > file - it's a user choice.
> >
> > The same isn't true of much of the code distributed
> > by Emacs.  And even when a distributed library (e.g.
> > `dired-x.el') is opt-in, Emacs should not suggest
> > bindings for its commands.  "Suggestion" by Emacs
> > is sometimes mistakenly taken by users as a "rule"
> > or a convention.
> 
> That's why it should be made clear that it's a suggestion, like:
> 
> "Many users find some commands not useful for them at all.  They might
> want to rebind their keys to ones that they use frequently."

That's not helpful, IMO.  Anyone can know that and
do that, without Emacs suggesting to bind specific
keys. And what one user finds not useful another one
finds useful.

(Why does "Many users..." remind me of DJT's "Many
people are saying..."? ;-))

But sure, many users find some things not useful for
them.  Many users aren't even aware of much of Emacs.
Most users, me included, use only a tiny bit of what
Emacs offers.

Users differ.  Use cases differ.  There are many ways
to use Emacs.

Any user who finds some key that is bound by default
not to be useful can rebind it.  That's not specific
to any particular key.  The doc should not be trying
to find and inform about keys that "many users" find
not so useful.

> > There's no good reason for Emacs to suggest that
> > users use `C-z' for anything particular.
> 
> On the contrary, there is: the meaning of C-z is "I want to leave Emacs
> for a moment and be able to come back".

Not in GUI Emacs, it's not, unless you consider
iconifying to be "leaving Emacs".