New Vim key bindings for Magit

[syl20bnr]

Commit 150e38e adds support for custom key bindings for Magit instead of the automatically defined evilified key bindings.

You can post your feedback here or in the official PR thread here: emacs-evil/evil-magit#2 (recommended)

[justbur]

I just noticed the new reset key, #, conflicts with magit-gh-pulls. I like # for reset, but I'm open to suggestions on how to resolve this.

[PierreR]

I believe it is really useful to be able to switch branches without going through the status buffer and pressing bb. So I have this in my .spacemacs.

(evil-leader/set-key "gc" 'magit-checkout)

Could we offer something similar in the standard spacemacs ?

[justbur]

@PierreR just do a PR for it. Seems reasonable to me. I think this issue was more about the decisions made within evil-magit.

[cpaulik]

Having

 (evil-snipe
  :variables
  evil-snipe-enable-alternate-f-and-t-behaviors t)

activated breaks the keybindings for F and f in magit

[TheBB]

For magit-gh-pulls and gitflow: well, @ and & are available, if you want to stick to those kinds of keys.

[justbur]

The single letter keys are mostly taken and are more of a pain to move. There are lots of modifier keys available and the "symbols". I like # for reset because of the analogy to gmail but C-r also works for me. I have no idea what the "natural keys" for gh-pulls and gitflow are. If reset is moved the old ones can stay.

[syl20bnr]

# is the official key binding of magit-gh-pull, would be better to not touch it. We have plenty of room for a reset binding that does not conflict.

[syl20bnr]

The single letter keys are mostly take and are more of a pain to move.

This is the price to pay for a custom evilification but it seems to worth it.

[syl20bnr]

C-r also works for me

I like it too. I would not take a Google key binding as a good reference in this context :-D
Otherwise @TheBB proposition of another symbol on the shifted number line could do it.

[justbur]

I think I like C-r the best, but I don't have a strong opinion

[justbur]

for what it's worth reset was on x

[syl20bnr]

C-x is not available and C-r is more mnemonic, looks good to me.

[justbur]

Ok. I made the change. Doesn't mean I can't change it again.

[syl20bnr]

If it changes it is for the better :-D

[herbertjones]

Does this mean evilification will be gone completely? Window bindings seem to be gone, if I M-1, instead of jumping to window 1, now it does a magit-section-show-level-1-all and a 2gt tries to switch to "Tracked files" instead of getting intercepted by the eyebrowse layer.

[justbur]

@herbertjones Those bindings were not provided by evilification. They just weren't being shadowed and now they are. It will be impossible to avoid all conflicts, so we'll just have to deal with the important cases. If you think "gt" for jump to tracked files is a bad binding, then propose an alternative. Otherwise we can disable that magit binding for the people who use the eyebrowse layer.

Same goes for the window select one, but it's worth pointing out that SPC 1 works fine

[herbertjones]

No, I'll just override them in my personal config as eyebrowse isn't a default layer. M-1 to M-9 are important to me as shells do not respect SPC-1 but do respect the Alt key.

[justbur]

We can easily override the M-[number] ones in spacemacs, depending on what
@syl20bnr thinks

On Fri, Oct 30, 2015 at 11:20 AM Herbert Jones notifications@github.com
wrote:

No, I'll just override them in my personal config as eyebrowse isn't a
default layer. M-1 to M-9 are important to me as shells do not respect
SPC-1 but do respect the Alt key.

—
Reply to this email directly or view it on GitHub
#3592 (comment)
.

[syl20bnr]

TBH, I'm really divided on these custom evil bindings, for now I want to believe that it worth the work put on them because Magit is such a central piece of the Emacs ecosystem that it could work.

But instead of stopping the leaks I would rather make it possible to easily choose between Magit custom evil key bindings and auto-evilified buffer (which will be my choice obviously :-)).

I see that the work to adapt the help popup of Magit could be applied to the auto-evilifed buffer as well.