Feature request: show key-binding when calling execute-extended-command with ivy enabled

[brabalan]

Ivy meets all my helm needs, while feeling faster, with one small exception: when I use M-x I don't see the binding of the command suggested. I have found this to be a great way to learn bindings … Would you consider adding this to ivy?

[abo-abo]

Would you consider adding this to ivy?

Sure, just need to write a separate function ivy-M-x, similar to helm-M-x. Right now, I'm using my fork of smex. I've sent the pull request ages ago, looks like it's not getting merged.

[brabalan]

Having an ivy-M-x would be great, but in the meantime I'll give your fork of smex a try. (A bit later). It's very nice indeed. I like how one can look-up key bindings, but it would be nice to see them displayed by default.

[abo-abo]

Have a look. Let me know what's missing.

[kaushalmodi]

Thanks for adding this.

Few comments from few minutes of trying it out:

• Can you sort the entries by last used instead of sorting the commands alpha-numerically? Currently the first command I see is 2C-associate-buffer which I probably am never going to use. It would be nice to see commands in the order of most recently used to least recently used.
• Is there a plan to enable fuzzy searching for certain commands using ivy/counsel? IMO counsel-M-x would be the first candidate to fit the bill for fuzzy searching. The reason is that many times the commands are long and it is faster to just ram in few letters from that whole command instead of remembering to hit a space after every few letters. Example: It is faster to type licdis instead of li c dis to get to list-colors-display.

[kaushalmodi]

It would be awesome to make C-. and C-, work in this command too (referring to your recent blog post: http://oremacs.com/2015/06/08/describe-variable-tip/).

[abo-abo]

Added smex sorting and extra bindings. I haven't looked to implementing own sorting so far, since it can be expensive to sort.

I'll look into fuzzy matching without a space soon.

[abo-abo]

Thanks, this fuzzy matcher isn't as slow as I thought it would be.
Actually it's as fast as the default for 8000 candidates.

[brabalan]

This is great, thanks a lot!

[abo-abo]

You're welcome.

[kaushalmodi]

The "recently used commands" cache stays stale until smex is called. No matter which command I execute using counsel-M-x, the first option in the command list stays the same, until I execute smex. So we also need to update the smex cache with each execution of counsel-M-x.

[kaushalmodi]

It was simply missing an smex-rank call:

(defun counsel-M-x (&optional initial-input)
  "Ivy version of `execute-extended-command'.
Optional INITIAL-INPUT is the initial input in the minibuffer."
  (interactive)
  (unless initial-input
    (setq initial-input (cdr (assoc this-command
                                    ivy-initial-inputs-alist))))
  (let ((ivy-format-function #'counsel--format-function-M-x)
        (cands obarray)
        (pred 'commandp)
        (sort t))
    (when (or (featurep 'smex)
              (package-installed-p 'smex))
      (require 'smex)
      (unless smex-initialized-p
        (smex-initialize))
      (smex-detect-new-commands)
      (smex-update)
      (setq cands smex-ido-cache)
      (setq pred nil)
      (setq sort nil))
    (ivy-read "M-x " cands
              :predicate pred
              :require-match t
              :history 'extended-command-history
              :action (lambda (cmd)
                        (smex-rank (intern cmd)) ; <-- FIX
                        (execute-extended-command current-prefix-arg cmd))
              :sort sort
              :keymap counsel-describe-map
              :initial-input initial-input)))

[abo-abo]

@kaushalmodi, thanks. But I think intern isn't necessary.