[mini.completion] "x" incorrectly added to React prop when accepting suggestion

[lukeramljak]

Contributing guidelines

• I have read CONTRIBUTING.md
• I have read CODE_OF_CONDUCT.md
• I have updated 'mini.nvim' to latest version of the main branch

Module(s)

mini.completion

Neovim version

0.12 (!at least latest Nightly build!)

Description

When completing a react prop name using mini.completion, an "x" character is sometimes inserted at the beginning of the prop.

Expected behaviour is that an "x" should not be inserted.

Reproduction

nvim version: v0.12.0-dev-1351+g8151fc59cf
vtsls version: 0.2.9

1. Clone the example repo with a minimal neovim config and sample react files: https://github.com/lukeramljak/nvim-completion-bug
2. Install the vtsls language server
3. Open react/App.tsx from the repo
4. Add the missing title prop to the component as per the screen recording below
5. Accept the completion suggestion for title using <c-y> or <cr>

Screen.Recording.2025-10-03.at.10.12.21.pm.mov

[echasnovski]

Thanks for the issue!

I'll try to reproduce and investigate during the weekend. There were several such issues lately and almost all of them ended up due to LSP server not fully following LSP specification (at least I think so). But I'll take a look.

[231tr0n]

All javascript frameworks related LSPs are raining down on mini.completion and mini.snippets :) . I uploaded a bug for svelte, another for vue and now finally react enters the chat :). And the funny part none of it might be related to mini.completion nor mini.snippets.

[echasnovski]

All javascript frameworks related LSPs are raining down on mini.completion and mini.snippets :) . I uploaded a bug for svelte, another for vue and now finally react enters the chat :). And the funny part none of it is related to mini.completion nor mini.snippets.

Yeah... If anything, this prompts to take a look if reverting the code addition or a safer approach is reasonable. But as far as I could infer, the reason is indeed the servers not following the spec. And discarding a useful feature because of that (and some internal Vim limitations) is sad.

[lukeramljak]

@echasnovski if that's the case, would you prefer if I created an issue for vtsls instead? I don't think you should have to remove features or make extra accommodations if it's caused by the LSP.

[echasnovski]

@echasnovski if that's the case, would you prefer if I created an issue for vtsls instead? I don't think you should have to remove features or make extra accommodations if it's caused by the LSP.

I'd have to first investigate whether it is indeed a server's issue.

It's just that the more servers pop up with this sort of issue the more likely that the issue is on 'mini.completion' / Neovim side. But to the best of my knowledge in all prior cases it was some issue on the server's side.