fix(repeatable_move): bug when nil was returned or normal command used; no longer requires { expr = true }

[mezdelex]

Should fix #775

2025-08-27.19-14-14.mp4

Not sure if this fix is already being addressed in another way or refactor process, but just in case, the move_repeatable function was failing when no string was being returned (nil in this case) due to { expr = true } keymap options.

[mezdelex]

I don't have a clue about Fennel, but it looks like there's a bug in the Fennel textobject parser

as I'm getting this error indicator (probably didn't install Fennel right...) but it's exactly at line 105 as the pipeline indicates

it looks like set_form text object doesn't have binding_pair

Anybody with more experience with Fennel I could ping? this looks like an existing bug

[kiyoon]

you can keep it commented so people don't set both keymaps by accident

[kiyoon]

Can you keep the change minimal, like get rid of the fennel textobject changes?

Also, you can just paste the snippet from the master branch because it's been working without issues. Introducing a new logic can introduce another bug.

[mezdelex]

@kiyoon oh yeah my apologies.

Discarded the mentioned textobject change and kept one of the both suggested keymaps options for repeatable_move commented as you suggested.

[kiyoon]

For example, the master has this snippet.

nvim-treesitter-textobjects/lua/nvim-treesitter/textobjects/repeatable_move.lua

Lines 107 to 121 in 71385f1

	if M.last_move.func == "f" or M.last_move.func == "t" then
	if opts.forward then
	vim.cmd([[normal! ]] .. vim.v.count1 .. ";")
	else
	vim.cmd([[normal! ]] .. vim.v.count1 .. ",")
	end
	elseif M.last_move.func == "F" or M.last_move.func == "T" then
	if opts.forward then
	vim.cmd([[normal! ]] .. vim.v.count1 .. ",")
	else
	vim.cmd([[normal! ]] .. vim.v.count1 .. ";")
	end
	else
	M.last_move.func(opts, unpack(M.last_move.additional_args))
	end

Is there any reason why you would change it to feedkeys? If not, I'd prefer to stick to the same logic.

[mezdelex]

No particular reason really, I just searched for a way to feed keys to the input, but the master branch snippet should in fact be more performant. Changed as suggested.