Doc that vi-mode requires "set -o vi" about fzf

[junegunn]

Ah, you're perfectly right, thanks for raising the issue. I was a bit worried about this too, but I haven't found a solution yet. I particularly don't like the fact that even if you have set -o vi in your .bashrc, if it comes after source ~/.fzf.bash, the key bindings will not be correctly set. Mentioning this quirkiness in the documentation would be helpful of course, and we should probably do so, but it would be better if we could fix the problem. Do you know of a better way to handle this?

[junegunn]

I've just updated the key bindings section of the README page. But I'm still interested in a real solution :)

[stephenh]

Hi Junegunn,

Thanks for updating the readme...good call about the "set -o vi" being before the source also being a gotcha.

I don't have any good ideas for a real solution.

Perhaps fzf could wait until it was invoked to detect vi mode? I vaguely recall seeing some arcane binding setup on Stackoverflow where invoking one binding would, via a bash function, conditionally setup/"bind" the next binding, and then itself automatically invoke that next binding.

I think the con was that it "burned" a binding, in that you had to use two (or three?) bindings to accomplish it. Plus it was fairly complex anyway.

[junegunn]

Thanks for the reply. I googled a bit and I guess you're referring to this, right?
http://stackoverflow.com/questions/8366450/complex-keybinding-in-bash

I roughly tested the approach as follows:

__fzf_rebind() {
  if [ -z "$(set -o | grep '^vi.*on')" ]; then
    bind '"\C-x\C-t": " \C-u \C-a\C-k$(__fsel_tmux)\e\C-e\C-y\C-a\C-d\C-y\ey\C-h"'
  else
    bind '"\C-x\C-t": "\e$a \eddi$(__fsel_tmux)\C-x\C-e\e0P$xa"'
  fi
}
bind -x '"\C-x\C-f": __fzf_rebind'
bind '"\C-t": "\C-x\C-f\C-x\C-t'

So it re-binds C-x + C-t to the right version depending on the current setting. It functionally works but when the rebind function is executed via C-x + C-f as an intermediate step, I see a line feed. I don't know if it's possible to suppress this unwanted line feed, but if it's not, we can't employ this approach.

If you don't mind, I'll keep this issue open until we find a clean solution to the problem.

[stephenh]

Hi Junegunn,

Yes, that was the article...nice find! Sorry for not including the link, I was on my other machine.

That is odd about the new line; I don't have any good ideas. I'll try it out later when I have some time.

Until then I agree leaving the issue open sounds like a good idea.

Thanks for the great support on these issues; I only just came across fzf but I think I will quickly be very addicted to it. It's a great idea.