tsi.el VS eglot

I've been building out tsi.el - an emacs minor mode which performs indentation based on a CST returned by tree-sitter. please give it a try if you haven't already, and if you have tried it and it did not meet your needs then I'd love to hear why not!

I'd really like to see something that combines the conciseness of SMIE's rule-based indentation system with tree-sitter's richer grammar. There are a couple of projects that I'm looking at, such as https://codeberg.org/FelipeLema/tree-sitter-indent.el and https://github.com/orzechowskid/tsi.el).

;; use our derived mode for tsx files (add-to-list 'auto-mode-alist '("\.tsx\'" . typescriptreact-mode)) ;; by default, typescript-mode is mapped to the treesitter typescript parser ;; use our derived mode to map .tsx -> typescriptreact-mode -> treesitter tsx (add-to-list 'tree-sitter-major-mode-language-alist '(typescriptreact-mode . tsx)) ;; https://github.com/orzechowskid/tsi.el/ ;; great tree-sitter-based indentation for typescript/tsx, css, json (quelpa '(tsi :fetcher github :repo "orzechowskid/tsi.el")) (add-hook 'typescript-mode-hook (lambda () (tsi-typescript-mode 1)))) ```

what are your specific problems? lsp, typescript-mode, and tree-sitter-mode together give me 90% of what I need. I fixed JSX/TSX indentation with a little tool I built (which I should really get around to publishing some day). I'm extremely happy with this setup and use it every day on professional and personal projects.

In this post I assume you already have some Emacs experience and now how to configure packages and write small Elisp snippets. I personally use straight.el as a package manager, minimal-emacs.d as a minimal vanilla Emacs configuration (along with my own custommizations), dape as the debug adapter client and eglot as my LSP client.

Recently I stumbled upon this issue:

https://github.com/joaotavora/eglot/discussions/1127

I don't know enough about emacs and LSP to see the full picture, but it seems that both eglot's and corfu's maintainers, assumably very competent programmers, can't find a solution for this.

I only skimmed the thread. My understanding is that LSP dumps a long list of completion candidates at once and they can't decide a cache strategy that works well with existing code...?

It could be. There are unfixed issues with eglot and corfu, and sadly not a lot of willingness to investigate.

Eglot errors when I add new Python code blocks. The error disappears when I reconnect the language server, but the same happens again when I add a new code block. My "workaround" now is that before I start working on the .qmd file, I just add a bunch of Python code blocks (for which I also have a function) and then reconnect the language server again. This way I can start working for a while until I need to add more code blocks again.

This discussion has helped with some performance issues: https://github.com/joaotavora/eglot/discussions/993.

I guess eglot doesn't support it yet: https://github.com/joaotavora/eglot/pull/839

This is not by pure chance, João is the developer of the Eglot LSP client and the breadcrumbs from LSP-mode had been requested as a feature, but as far as I remember João thought rightfully that this could be an independent package, see https://github.com/joaotavora/eglot/discussions/988

A substantial section of the community is using corfu instead of company, but I wouldn't say company is out of date by any means. In emacs 29 eglot will be a built in, which might act as a replacement for lsp-mode depending on what functionality you need.

I am currently running emacs 29 (built from emacs-29 branch) which – according to https://github.com/joaotavora/eglot – should contain the latest eglot.

Yep. You can use flymake or flycheck for that in combination with eglot or lsp-mode.

See https://github.com/joaotavora/eglot#diagnostics