feat: add common default keybindings

[stevenxxiu]

For [Feature Request] Custom keyboard shortcuts · Issue #610 · ellie/atuin.

[trygveaa]

I think it would be beneficial if this behaved in the same way as bash and zsh. The differences I see is that they always move to the beginning/end of the next word with ctrl-left/right. So with e.g. test()-test and the cursor at the end, ctrl-left first moves after - and then to the beginning. And then ctrl-right moves before ( and then to the end. ctrl-delete and alt-backspace does the same movements, but deletes the words instead (ctrl-backspace doesn't do anything in me for bash or zsh, and just moves the cursor in this branch for me).

Another difference is that _ is counted as a word separator in bash and zsh (I haven't checked all other symbols).

(I'm not affiliated with atuin btw, so these are just my personal opinions).

I also notice that if I press ctrl-right or ctrl-delete on an empty prompt, atuin panics with this branch.

[stevenxxiu]

I also notice that if I press ctrl-right or ctrl-delete on an empty prompt, atuin panics with this branch.

Thanks for testing. Should be fixed now. I added a test case for that.

I think it would be beneficial if this behaved in the same way as bash and zsh.

I made the word jumping action identical to Sublime Text, which is what I'm used to. In fact, I configured my Zsh line editor to behave the same way. I use vi-backward-word and vi-forward-word in my configuration with a custom WORDCHARS variable. (I'm thinking about switching to Nushell soon.)

I tried your example of test()-test. The current behavior also matches my browser (Vivaldi).

I think word jumping behavior is up to personal preference. It makes sense to add some configuration option for this. Would a configurable WORD_SEPARATORS suit?

[trygveaa]

Ah, right, emacs-forward/backward-word and vi-forward/backward-word behave differently in zsh and none of them are the default. I think bash behaves the way I said by default though. I think it would make sense to stick close to the default behavior of common shells, since this is a tool for shell history. I'll defer to the maintainers here to decide this and to decide whether a configuration option should be added.

[conradludgate]

We already have a mechanism for deleting words: Ctrl-W. I've just tested it and it treats test()-test as a single word.

As I don't use this feature in any shell and I'm not an emacs/vim keybinding user I can't really comment on which behaviour works best here

[ellie]

Ah, right, emacs-forward/backward-word and vi-forward/backward-word behave differently in zsh and none of them are the default. I think bash behaves the way I said by default though. I think it would make sense to stick close to the default behavior of common shells, since this is a tool for shell history. I'll defer to the maintainers here to decide this and to decide whether a configuration option should be added.

There doesn't really seem to be an obvious answer, so I think going for the default bash behaviour is the best option - I imagine most people will be used to this!

Otherwise, making things configurable would be 💯

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated
atuin	✅ Ready (Inspect)	Visit Preview	💬 Add your feedback	Mar 1, 2023 at 7:56AM (UTC)

[stevenxxiu]

I added configurable word_jump_mode (emacs/subl for Sublime Text) and word_chars options now.

Hope this matches your expectations of Emacs behavior. Please let me know if it's still incorrect in some way.

[stevenxxiu]

@ellie Sorry about the nag but have time to look at this?

[ellie]

I'm happy for this to go in, thank you for the work! 🚀 And apologies for the slower review :/