How do I get rid of RET being translated from <return>?

Question

I'm overhauling all of my keybinds and I'm trying to keep it logically consistent.

When I issue the command describe-key -> RET
This pops up in the *Help* buffer:

RET (translated from <return>) runs the command newline (found in global-map),
which is an interactive compiled Lisp function in ‘simple.el’.

It is bound to RET.

(newline &optional ARG INTERACTIVE)

What I don't like about this is the fact that RET is translated from < return >.
This is also the case with DEL -> < backspace >, M-DEL -> < M-backspace >, etc.

So to get rid of said indirections I issue the command:
(define-key function-key-map (kbd "<return>") 'nil)

And thus bind RET to newline:
(define-key global-map (kbd "RET") 'newline)

But then I get this error in the minibuffer when pressing the RET key:
<return> is undefined

How do I get rid of this behavior?

I tried reading the documentation on map translations, but it is very confusing to me. Perhaps I haven't had that AHA moment.

Answer 1

For clarity: "RET (translated from <return>)" tells you that Emacs is receiving <return>. That's what your OS sent it when you typed that key. You want Emacs to translate <return> to RET for you, because the standard key bindings are for RET, not <return>. Hence the error "<return> is undefined" after you removed that translation.

The reason for all this is that terminals send RET (which is ASCII code 13, aka C-m or ^M), whereas in a windowing/GUI system the return/enter key has its own distinct code <return> (which terminals don't know about, and can't send). Because Emacs supports both, it makes sense for the key bindings to be for RET.

Personally I recommend that you just leave things the way they are; but if you can't, then maybe something like this:

(define-key special-event-map (kbd "<return>") 'my-ret-event)

(defun my-ret-event ()
  "Push RET onto `unread-command-events'."
  (interactive)
  (push '(t . ?\C-m) unread-command-events))

Refer to C-hig (elisp)Special Events

Any key bindings which are actually using <return> then become inaccessible (that's probably an empty list, as such bindings wouldn't work in a terminal, but obviously it's still possible).

Experimentally, the various translation and decoding maps (although I didn't check every option) all seem cause Emacs to report the translation; but the special events are processed further behind the scenes.

n.b. I'm not convinced this is sensible, and for all I know it might cause problems in some situations.

Someone else may have a better method.