Comments on: Emacs + Evil = ecumenicalism

By: maindoor

maindoor — Mon, 17 Sep 2012 08:06:41 +0000

I too tried evil, recently. Found it buggy. Definitely better
compared to Viper/Vimpulse, but still buggy. simple eg.
like copying to registers doesn't work. "n2yy. Still needs
those final touches.

By: Emacs Bits and Bobs | Simon's Coding Blog

Emacs Bits and Bobs | Simon's Coding Blog — Thu, 13 Sep 2012 13:44:45 +0000

[...] done in my config. I’ve cobbled alot of this from various places on the intermawebs, notably http://dnquark.com/blog/2012/02/emacs-evil-ecumenicalism/ and https://github.com/cofi/dotfiles/blob/master/emacs.d/config/cofi-evil.el ;;EVIL stuff (require [...]

By: Lele

Lele — Sun, 22 Jul 2012 02:04:13 +0000

I tried Evil, recently. I found it buggy. I'm sticking to Vimpulse for the time being. Yes, Vimpulse is discontinued, but it gives you enough Vim power to be dangerous. Thank you for sharing, anyway. The future is Evil, anyway. Cheers.

By: Petro

Petro — Tue, 15 May 2012 18:19:06 +0000

Great.
I just discovered evil mode. One more thing I can recommend is Pentadactyl addon for firefox.

By: chen bin

chen bin — Sun, 04 Mar 2012 13:36:44 +0000

This is the best post on evil-mode. Besides, I love your post on org2blog.

By: Frank Fischer

Frank Fischer — Thu, 01 Mar 2012 15:12:11 +0000

You may try the following motions for sexp-navigation. The main difference to the Emacs ones is that (point) is that the closing ")" is considered as the end of an sexp and point is placed there instead of the following character. Note that this is only a quick hack and widely untested (there may very well be edge cases)!

(evil-define-motion evil-fwd-sexp (count)
:type inclusive
(setq count (or count 1))
(unless (zerop count)
(goto-char
(save-excursion
(condition-case err
(let ((opoint (point)))
(cond
((> count 0)
(condition-case nil
(forward-sexp)
(error (forward-char)))
(when (> (point) (1+ opoint))
(setq count (1- count)))
(forward-sexp count)
(backward-char)
(point))
(t
(setq count (- count))
(forward-char)
(condition-case nil
(progn
(backward-sexp)
(setq count (1- count)))
(error (backward-char)))
(backward-sexp count)
(point)))))))))

(evil-define-motion evil-bwd-sexp (count)
:type inclusive
(evil-fwd-sexp (- (or count 1))))

By: Michael Markert

Michael Markert — Wed, 22 Feb 2012 14:50:32 +0000

Thanks for your insightful post, great that there are some people who come from Emacs to modal editing via Evil :)

Some points to add:

@Combining Commands
vfa; can also be written as v2fa just to demonstrate that f(ind) and t(o) also take numeric arguments (think of M-2 for the rest of Emacs). Also most other (motion) commands take numeric arguments.

Evil supports text objects so you can also define your di$KEY to delete inside such a text object. (For more see evil-commands.el)

@Defining normal mode commands
From the Merge Request emerged the evil-leader package, to be found here: https://github.com/cofi/evil-leader (I also added it now to the plugins list on emacswiki/Evil)

@My dotfiles
cofi-evil.el depends on cofi-util.el and the cl package, if you copied code and something fails look there ;) (well and the other `require'd packages)

evil-{define,declare}-key support also more than one definition at a time to keep your key bindings clean.

I mapped my CapsLock to Ctrl and use `jk' to exit insert mode (look for cofi/evil-maybe-exit) because this way using other Emacs packages (that sadly never heard about Evil) is way easier. You can also exit insert state via C-[ but that's quite cumbersome.

Last but not least: Evil is discussed at the vim-emulation mailing list also accessible via gmane: http://news.gmane.org/gmane.emacs.vim-emulation that surely should also go into the 'Where you can learn more' list ;)

By: dnquark

dnquark — Wed, 22 Feb 2012 13:07:38 +0000

@eric: Actually, the point that you make is pretty important, and better captures the Emacs/Vim selection behavior difference than my description. Also, IIUC, "between characters" is also the correct way to think of the cursor positioning in vi insert mode, while "on character" is the visual/normal state behavior. So perhaps the asymmetry between Emacs and vi is really a consequence of that "internal asymmetry" in vi itself. The reason it's important to think about this is that using some Emacs motion commands while in Evil normal/visual mode results in unexpected or incorrect (usually, in the off-by-one sense) behavior due to the different underlying model of the cursor positioning. Sexp-motion commands are the key culprits here, but there might be others. Ideally, I'd like to see Evil compensating for this difference in behavior (see this discussion), but it's somewhat subtle and I personally don't have a clear conception of how one would do this in such a way as to preserve behavior that's sensible for Emacs without breaking Vim commands. (Alternatively, someone could tell me how to navigate through code in Evil effectively without resorting to sexp-motions. I have found them to be very handy, and I know seasoned Emacs users that swear by them for any sort of code navigation.)

By: eric

eric — Wed, 22 Feb 2012 11:30:11 +0000

I am a long time (very long time) user of Emacs who learned vi on Unix V7 in ancient times and had always pined for a modal editor. Evil has been a fantastic development, managing to combine modal editing with Emacs in a way that other attempts hadn't managed. Your blog is a good description of the issues that do arise and I enjoyed reading it.

One minor possibly pedantic point: I would argue that there is no asymmetry in how Emacs handles selections. The key difference between Emacs and vi is that in emacs the position is always "between" characters where in vi it is "on" a character. For this reason, I have always set the cursor in Emacs to be a vertical bar to highlight this aspect. When using evil, I keep the cursor as a block. The difference positioning between the modes is then made visually clear.

Thanks again!

By: Arne Babenhauserheide

Arne Babenhauserheide — Tue, 21 Feb 2012 20:06:32 +0000

I use control-lock-mode to achieve the same, without sacrifizing all the knowledge of useful key-combinations.