Customizing universal-argument (C-u), instead of hardcoding 4?

Customizing universal-argument (C-u), instead of hardcoding 4?

[Karl Fogel]

This question has surely come up here before, but I searched and didn't find the discussion.

The default value of 4 for `universal-argument' (the C-u prefix) is currently hardcoded.  It could be useful for it to be customizable.  One example: I find that the distance 'C-u C-p' and 'C-u C-n' move by default is a bit too small for most of my use cases, and often the multiples are not quite right either.  I'm trying some other values, especially 5, and thinking "Hmm, why not make it easy for people to experiment and set the right value for them?"

Of course the default C-u value affects more than C-p and C-n, but the question is the same: why hardcode it?  There's nothing special about 4, and we don't really know if it's the right value for most people.

It's set in simple.el, and there are various places in the code that expect it to be 4 as well.  I haven't looked deeply into how hard it would be to change, because I first wanted to ask here if a) this has been considered, and b) if anyone would be opposed to customizability & if so why.  But here's what I found in a quick look at the code:

These are the core places in simple.el that would need to refer to a new customizable variable instead of hardcoding 4 (I don't think any modification to prefix-numeric-value() in callint.c is necessary, as it gets the value from Lisp):

  ;; FIXME: Declaration of `prefix-arg' should be moved here!?
  
  (add-hook 'prefix-command-echo-keystrokes-functions
            #'universal-argument--description)
  (defun universal-argument--description ()
    (when prefix-arg
      (concat "C-u"
              (pcase prefix-arg
                (`(-) " -")
                (`(,(and (pred integerp) n))
                 (let ((str ""))
                   (while (and (> n 4) (= (mod n 4) 0))
                     (setq str (concat str " C-u"))
                     (setq n (/ n 4)))
                   (if (= n 4) str (format " %s" prefix-arg))))
                (_ (format " %s" prefix-arg))))))
  
  
  (defun universal-argument ()
    "Begin a numeric argument for the following command.
  Digits or minus sign following \\[universal-argument] make up the numeric > argument.
  \\[universal-argument] following the digits or minus sign ends the argument.
  \\[universal-argument] without digits or minus sign provides 4 as argument.
  Repeating \\[universal-argument] without digits or minus sign
   multiplies the argument by 4 each time.
  For some commands, just \\[universal-argument] by itself serves as a flag
  which is different in effect from any particular numeric argument.
  These commands include \\[set-mark-command] and \\[start-kbd-macro]."
    (interactive)
    (prefix-command-preserve-state)
    (setq prefix-arg (list 4))
    (universal-argument--mode))
  
  (defun universal-argument-more (arg)
    ;; A subsequent C-u means to multiply the factor by 4 if we've typed
    ;; nothing but C-u's; otherwise it means to terminate the prefix arg.
    (interactive "P")
    (prefix-command-preserve-state)
    (setq prefix-arg (if (consp arg)
                         (list (* 4 (car arg)))
                       (if (eq arg '-)
                           (list -4)
                         arg)))
    (when (consp prefix-arg) (universal-argument--mode)))

And here are some places in the code that currently seem to rely on the value being 4 or some power thereof, so we'd have to adjust them to to look at a variable (in some cases raised to a power >1):

  ./lisp/simple.el:   ((and (consp arg) (> (prefix-numeric-value arg) 4))
  ./lisp/vc/pcvs.el:	 (new (> (prefix-numeric-value current-prefix-arg) 8))
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)))
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)
  ./lisp/vc/diff-mode.el:      (when (> (prefix-numeric-value other-file) 8)
  ./lisp/progmodes/sql.el:      (when (>= (prefix-numeric-value product) 16)
  ./lisp/progmodes/sql.el:         ((= (prefix-numeric-value product) 4) ; C-u, prompt for product
  ./lisp/progmodes/python.el:        (= (prefix-numeric-value current-prefix-arg) 4))
  ./lisp/font-lock.el:		(let ((lines (if arg (prefix-numeric-value arg) 16)))
  ./lisp/allout.el:                             ((<= (prefix-numeric-value keymode-cue) 4)
  ./lisp/allout.el:                             ((> (prefix-numeric-value keymode-cue) 4)
  ./lisp/emulation/cua-base.el:   ((and (consp arg) (> (prefix-numeric-value arg) 4))
  ./lisp/dabbrev.el:	  (and arg (= (prefix-numeric-value arg) 16)))
  ./lisp/dired.el:	 (if current-prefix-arg ?\040)))
  ./lisp/dired.el:	 (if current-prefix-arg ?\040)))
  ./lisp/simple.el:		    (eq (car current-prefix-arg) 4)) "C-u ")
  ./lisp/vc/pcvs.el:	 (new (> (prefix-numeric-value current-prefix-arg) 8))
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)))
  ./lisp/vc/pcvs.el:	      (> (prefix-numeric-value current-prefix-arg) 8)
  ./lisp/vc/pcvs.el:	     (let ((current-prefix-arg '(4)))
  ./lisp/vc/pcvs.el:    (let ((current-prefix-arg '(16)))
  ./lisp/vc/vc-git.el:      ((equal current-prefix-arg '(16))
  ./lisp/vc/vc-git.el:	  (if (equal current-prefix-arg '(4))
  ./lisp/progmodes/grep.el:      ((and grep-command (equal current-prefix-arg '(16)))
  ./lisp/progmodes/grep.el:		(confirm (equal current-prefix-arg '(4))))
  ./lisp/progmodes/grep.el:      ((and grep-find-command (equal current-prefix-arg '(16)))
  ./lisp/progmodes/grep.el:		(confirm (equal current-prefix-arg '(4))))
  ./lisp/progmodes/grep.el:	((and grep-find-command (equal current-prefix-arg '(16)))
  ./lisp/progmodes/grep.el:		  (confirm (equal current-prefix-arg '(4))))
  ./lisp/progmodes/python.el:        (= (prefix-numeric-value current-prefix-arg) 4))
  ./lisp/org/org.el:	     (if (and (not (equal current-prefix-arg '(4)))
  ./lisp/sort.el:	   (equal current-prefix-arg '(4))
  ./lisp/sort.el:	   (equal current-prefix-arg '(16))
  ./lisp/sort.el:	   (equal current-prefix-arg '(64))

(Again, this was a very cursory glance, so I'm sure I missed spots.)

Thoughts?

Best regards,
-Karl

Re: Customizing universal-argument (C-u), instead of hardcoding 4?

[John Wiegley]

>>>>> Karl Fogel <kfogel@red-bean.com> writes:

> It's set in simple.el, and there are various places in the code that expect
> it to be 4 as well. I haven't looked deeply into how hard it would be to
> change, because I first wanted to ask here if a) this has been considered,
> and b) if anyone would be opposed to customizability & if so why. But here's
> what I found in a quick look at the code:

I'll leave it to others to answer (a).

As for (b), I am opposed to the change. There is too much expectation right
now that it has the value 4, including code that does not live in the Emacs or
GNU ELPA repositories.

If your main reason is to make C-u movement customizable, how about rebinding
C-u to "my-universal-argument"?

-- 
John Wiegley                  GPG fingerprint = 4710 CF98 AF9B 327B B80F
http://newartisans.com                          60E1 46C4 BD1A 7AC1 4BA2

Re: Customizing universal-argument (C-u), instead of hardcoding 4?

[Karl Fogel]

John Wiegley <jwiegley@gmail.com> writes:
>I'll leave it to others to answer (a).
>
>As for (b), I am opposed to the change. There is too much expectation right
>now that it has the value 4, including code that does not live in the Emacs or
>GNU ELPA repositories.
>
>If your main reason is to make C-u movement customizable, how about rebinding
>C-u to "my-universal-argument"?

Thanks, John.  This wasn't something I feel strongly about, so if you're opposed, that's enough for me to withdraw the suggestion.  I can roll my own `universal-argument and `universal-argument-more' if I really want to; obviously it's easier (for me) the less code I maintain outside the upstream sources, but it's no big deal.  I hadn't thought about the point that lots of code outside Emacs and even ELPA is probably also counting on it being 4, but now that you mention that, it makes me kind of take your side rather than mine anyway :-).

Best regards,
-Karl

Re: Customizing universal-argument (C-u), instead of hardcoding 4?

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

Each command can distinguish a C-u from an explicit argument of 4.
So if you want to make C-n and C-p move 5 instead, you can do it
in those commands without changing anything else.

-- 
Dr Richard Stallman
President, Free Software Foundation (gnu.org, fsf.org)
Internet Hall-of-Famer (internethalloffame.org)
Skype: No way! See stallman.org/skype.html.