Beta-testing 'mini.keymap'

[echasnovski]

Contributing guidelines

• I have read CONTRIBUTING.md
• I have read CODE_OF_CONDUCT.md

Module(s)

mini.keymap

Description

Please leave your feedback about new mini.keymap module here. Feel free to either add new comment or positively upvote existing one.

Some things I am interested to find out (obviously, besides bugs):

• What built-in steps for map_multistep() that are currently missing would you find useful? They probably have to be convenient to use as one of <Tab> / <S-Tab> / <CR> / <BS> key in Insert mode.
• Is configuration of common mappings intuitive enough?
• Is documentation and examples clear enough?

Thanks!

[pkazmier]

I like the sample kk mapping to correct the last spelling mistake. That's a nice idea. How would I modify that so it doesn't automatically select the first option, but instead allow me to choose from the menu, and then resume the gi part?

BTW, I've already fallen in love with <Tab> bound to jump_after_tsnode and jump_after_close and the reverse for <S-Tab>. Great stuff!

[echasnovski]

I like the sample kk mapping to correct the last spelling mistake. That's a nice idea. How would I modify that so it doesn't automatically select the first option, but instead allow me to choose from the menu, and then resume the gi part?

I can't think of anything more appropriate than <BS><BS><Esc>[sz= followed by choosing a candidate and manual gi<Right>.

If you use 'mini.pick' (which I hope you still do, despite your great interest in 'snacks.picker' :) ), I think writing a custom function that uses MiniExtra.pickers.spellsuggest picker followed by vim.api.nvim_input('gi<Right>') should be doable (because 'mini.pick' picker waits for the picker to finish before processing code further). Probably doable with other picker too, but needs tweaking "choose" action.

BTW, I've already fallen in love with <Tab> bound to jump_after_tsnode and jump_after_close and the reverse for <S-Tab>. Great stuff!

Thanks! I can't take credit for the idea as "tabout" is a feature with history in several editors. But it indeed fits nicely into the module scope.

[drowning-cat]

TL;DR

• Contextual multi-purpose keymaps
• Wait-free keymaps

[echasnovski]

TL;DR

* Contextual multi-purpose keymaps

* Wait-free keymaps

Something like that, yes. The first one is more or less accurate (mostly because "context" is usually something depending on text state, while it can be arbitrary per-step condition). The second one - not so much, as they are not really mappings and there is already <nowait> flag in regular mappings.

[drowning-cat]

I'm starting to get it a little.

If I understand correctly, map_combo keymaps behave almost the same way as <nowait> keymaps but allow for controlling the exact delay per mapping.

vim.keymap.set('i', 'kl', 'kl<Esc>', { nowait = true })
vim.keymap.set('i', 'kkl', 'kkl<Esc>', { nowait = true })
--
local map_combo = require('mini.keymap').map_combo
map_combo('i', 'kl', '<Esc>', { delay = 75 })
map_combo('i', 'kkl', '<Esc>', { delay = 75 })

[abeldekat]

I very much like the new module!

Having learned my lessons with hardtime I can now get by with only a check on jjj... and kkk......)

The smart snippet mappings are also useful when not using tab:

local mappings = nil
if MiniKeymap then
  local map_multistep = MiniKeymap.map_multistep
  local next_or_out = { "minisnippets_next", "jump_after_tsnode", "jump_after_close" }
  map_multistep("i", "<C-l>", next_or_out)

  local prev_or_out = { "minisnippets_prev", "jump_before_tsnode", "jump_before_open" }
  map_multistep("i", "<C-h>", prev_or_out)

  mappings = { jump_next = "", jump_prev = "" } -- already defined above...
end

mini_snippets.setup({
  -- ...
  mappings = mappings,
  -- ...
})

[echasnovski]

If I understand correctly, map_combo keymaps behave the same way as <nowait> keymaps but allow for controlling the exact delay per mapping.

Not quite. The whole reason for this "better escape" like approaches to mappings is the behavior of the cursor when typing initial keys from the left hand side:

• With regular mappings there is a visible visual delay for a duration of 'timeoutlen' during which Neovim waits for the next key.
• With combos (and "better escape" like approaches) there is no such side effects.

Plus, there can be several combos for the same LHS.

[echasnovski]

The smart snippet mappings are also useful when not using tab:

Yeah, that's also possible of course.

---

By the way, one small thing from behind the curtain. I initially planned to have dedicated key-scoped functions (like map_multi_tab(), map_multi_shift_tab(), etc.), but seeing your usage of <C-k> in #1767 made me change my mind. And it was a good decision, I feel. So thank you for that :)

[echaya]

As always, thanks for your amazing mini series. I'm a long time better-escape user and cant wait to move the config to the family :)

Below is what I have configured:

local map_combo = require('mini.keymap').map_combo
local mode = { 'i', 'c', 'x', 's' }
map_combo(mode, 'jk', '<BS><BS><Esc>')
map_combo('t', 'jk', '<BS><BS><C-\\><C-n>')
map_combo('n', 'jk', ':wincmd =<cr>k')

Issues I see (in both windows, wsl) within wezterm:

1. The cursor shape doesn't always update when jump out from insert mode
2. jk to jump out from terminal mode doesn't work unless I jump out manually using <c-> for once
3. I haven't got map_combo working consistently on normal mode.

Separately, on better-escape I set timeout=150, wonder if we can setup something similar here; I use better-escape for normal mode keymap (jk) so that it will only be triggered if I typed the combo fast (within 150ms)

[echasnovski]

• The cursor shape doesn't always update when jump out from insert mode

• jk to jump out from terminal mode doesn't work unless I jump out manually using for once

• I haven't got map_combo working consistently on normal mode.

Sorry, I can not reproduce any of the issues described. Everything works as expected here. And it is a bit hard to see in the demo because there is no indication of either mode or pressed keys. Are you sure there is nothing conflicting with these mappings? Maybe 'better-escape' is still applied somehow? The best way to be sure is to reproduce with only 'mini.keymap' mappings enabled.

One thing I was able to sometimes somewhat see is that pressing jk in Terminal mode (with map_combo('t', 'jk', '<BS><BS><C-\\><C-n>') setup) doesn't always redraw statusline mode. As it is not reliable and only in Terminal mode, not sure what to do here.

Separately, on better-escape I set timeout=150, wonder if we can setup something similar here; I use better-escape for normal mode keymap (jk) so that it will only be triggered if I typed the combo fast (within 150ms)

It is mentioned both in examples and map_combo() documentation: set { delay = 150 } as fourth argument to adjust delay for specific combos.

[echaya]

• The cursor shape doesn't always update when jump out from insert mode
• jk to jump out from terminal mode doesn't work unless I jump out manually using for once
• I haven't got map_combo working consistently on normal mode.

Sorry, I can not reproduce any of the issues described. Everything works as expected here. And it is a bit hard to see in the demo because there is no indication of either mode or pressed keys. Are you sure there is nothing conflicting with these mappings? Maybe 'better-escape' is still applied somehow? The best way to be sure is to reproduce with only 'mini.keymap' mappings enabled.

One thing I was able to sometimes somewhat see is that pressing jk in Terminal mode (with map_combo('t', 'jk', '<BS><BS><C-\\><C-n>') setup) doesn't always redraw statusline mode. As it is not reliable and only in Terminal mode, not sure what to do here.

Separately, on better-escape I set timeout=150, wonder if we can setup something similar here; I use better-escape for normal mode keymap (jk) so that it will only be triggered if I typed the combo fast (within 150ms)

It is mentioned both in examples and map_combo() documentation: set { delay = 150 } as fourth argument to adjust delay for specific combos.

Tested again with {delay=150} added to the keymap (I have removed better-escape from loading)

• The cursor shape doesn't always update when jump out from insert mode => the issue can be seen in WSL (using wezterm or Windows terminal using nvim 0.10.1). Still persists after upudpate nvim to 0.11.1. Can be consistently replicated within the following line when cursor is at the last letter of the line:
  local map_combo = require("mini.keymap").map_combo

• jk to jump out from terminal mode doesn't work unless I jump out manually using for once => the issue can be consistently replicated with snacks_terminal for both windows and WSL but not from native terminal launched from cmdline

• I haven't got map_combo working consistently on normal mode => the issue is seen in nvim 0.10.1 but after update nvim to 0.11.1 it is gone.

[echasnovski]

• The cursor shape doesn't always update when jump out from insert mode => the issue can be seen in WSL (using wezterm or Windows terminal using nvim 0.10.1). Still persists after upudpate nvim to 0.11.1. Can be consistently replicated within the following line when cursor is at the last letter of the line:
  local map_combo = require("mini.keymap").map_combo

I can't think of any reason for that, unfortunately. Maybe a WSL thing. Here is how it works for me:

mini-keymap_working-jk.mp4

• jk to jump out from terminal mode doesn't work unless I jump out manually using for once => the issue can be consistently replicated with snacks_terminal for both windows and WSL but not from native terminal launched from cmdline

Then it is something to do with 'snacks.terminal'. Maybe you need to adjust RHS of the combo here.

• I haven't got map_combo working consistently on normal mode => the issue is seen in nvim 0.10.1 but after update nvim to 0.11.1 it is gone.

That's good enough. Might also try '<Cmd>wincmd =<CR>k' as a right hand side on earlier versions.