Any objection to adding completing-read-function?

[Leo]

Hello all,

I have been locally using a variable like that to replace
completing-read with ido-completing-read and it appears to me to improve
efficiency in many places.

So I wonder if there is any objection to adding a new variable
completing-read-function that when set replaces completing-read? Let me
know if I should submit it to the bug tracker.

After applying the attached patch, one may customise it like this:

(setq completing-read-function 'ido-completing-read*)
(defun ido-completing-read* (prompt choices &optional predicate require-match
                                    initial-input hist def inherit-input-method)
  (if (and (listp choices) (not (functionp choices)))
      (ido-completing-read prompt (if (listp (car choices))
                                      (mapcar 'car choices)
                                    choices)
                           predicate require-match initial-input hist def 
inherit-input-method)
    (let ((completing-read-function nil))
      (completing-read prompt choices predicate require-match
                       initial-input hist def inherit-input-method))))

Here are some screenshots:

I switching to a bookmark:

bookmark.png
Description: PNG image

Switching to a branch in magit:

emacs.png
Description: PNG image

Loading lisp systems in slime:

slime.png
Description: PNG image

Patch:

crf.diff
Description: Text Data

Cheers,
Leo