fix: schedule vim commands in state_spec test

[seflue]

Fix test on CLI AND Neotest by executing the vim commands asynchronously.

This test was not marked as failed when it ran from the command line, although the error was visible in the output.
In Neotest it was actually marked as failed.

[kristijanhusak]

although the error was visible in the output

I can't see any errors in the CLI output:

╰─ make test FILE=tests/plenary/state/state_spec.lua                                                                                                                                                                                         ─╯
zsh: correct 'test' to 'tests' [nyae]? n
nvim --headless --clean -u tests/test.lua "tests/plenary/state/state_spec.lua"
Individual Test File/Directory provided: tests/plenary/state/state_spec.lua
Running all tests at tests/plenary/state/state_spec.lua
Starting...Scheduling: tests/plenary/state/state_spec.lua

========================================	
Testing: 	/home/kristijan/github/orgmode/tests/plenary/state/state_spec.lua	
Success	||	State should create a state file if it doesn't exist	
Success	||	State should save the cache file as valid json	
Success	||	State should be able to save and load state data	
Success	||	State should be able to self-heal from an invalid state file	
	
Success: 	4	
Failed : 	0	
Errors : 	0	
========================================	

Can you provide some info what you get?

[seflue]

Within neotest I get the following error:

On CLI I actually also had 0 errors, regarding the output I might have seen a gost (a couple of tests throw some output and it's quite possible, that I misread it).
To check against neotest, you would need my last state of neotest_plenary with some funky reconfiguration:

Or you rename the scripts/minimal_init.lua into something which isn't catched by the neovim-plenary globbing.
Background: neovim-plenary is globbing for a minimal_init.lua, and tests/minimal_init.lua in the orgmode project is the correct one. But because scripts/minimal_init.lua matches first, so the tests don't run at all with neotest. My fix makes the globbing configurable and also fixes the configurability of neotest-plenary in a general sense. Being able to run the tests from within neovim is much more convenient during development.

[kristijanhusak]

Can you try these changes instead and see if it works?

Suggested change

	vim.schedule_wrap(function()
	vim.cmd.edit(cache_path)
	vim.api.nvim_buf_set_lines(0, 0, -1, false, { '[ invalid json!' })
	vim.cmd.write()
	end)
	vim.cmd(('edit %s'):format(cache_path))
	vim.api.nvim_buf_set_lines(0, 0, -1, false, { '[ invalid json!' })
	vim.cmd('write')

From the screenshot I see that neotest-plenary cannot figure out that vim.cmd is a table, so this might fix it.

[seflue]

Thanks, this actually also fixes the problem. I updated my commit with your suggestion.