tabnine-vscode VS lsp-mode

Both are related to limitations of vscode completion engine, so Tabnine can't really blamed for them in any way.

1. Initially Tabnine's auto-completion was triggered on any character, which best leveraged Tabnine's power but also had inherent problem: when Tabnine was triggered on non-letter character it sometimes prevented Vscode from showing suggestions from other completion sources (LSPs, snippets). There is a discussion in https://github.com/codota/tabnine-vscode/issues/6 with me explaining that the only viable solution is to reduce set of trigger characters to letters only. In the end a fix was pushed that reduced the set of trigger characters, which made the problem less likely but still not solved. The are numerous duplicates of this issue on Github.

2. Another problem is when Vscode has auto-completion suggestions from Tabnine and other sources (LSP, snippets), it frequently puts Tabnine's at the top of list. This is a big no-go for me because most of the time just want to complete the identifier (class field, method etc.). Modifying just the extension code didn't help so in the end I had add a small patch to Vscode itself, which gives lowest score to Tabnine's candidates: https://github.com/sergei-dyshel/vscode/commit/ee73034b9ec6c....

I must admit that both problems can be practically solved by new "inline auto-completion" mechanism in vscode which looks very promising for AI-based completion in general. I'm looking forward to evaluate it.

I also tried to the solutions suggested  and , but nothing worked. I moved the (setq lsp-keymap-...) line outside (and before) use-package. I also used :config (define-key lsp-load-map...) in my use-package block. But none of them worked.

Not sure how much this might help, but lsp-mode has lsp-yaml-select-buffer-schema and lsp-yaml-set-buffer-schema commands to pick schema from a list or set from a URI. Checking the source of them might give some hints about how the same could be implemented in eglot?

Beyond that you might as well embrace the suck and install autex with a language server: https://emacs-lsp.github.io/lsp-mode/

Smart completion these days is done primarily through LSP. eglot is fairly minimal but built-in as of 29, also available via GNU Elpa. lsp-mode is another option with more integrations and a bit more fleshed out.

lsp-mode: https://github.com/emacs-lsp/lsp-mode/issues/2080

Are you running emacs-29? It has numerous speed-ups compared to emacs-28 and older versions, many of them coded by Mattias Engdegård, e.g. commit def6fa4246. I have a fresh build of emacs-29 running on Linux and a new mac with an M1 CPU, and it's stupid fast. I don't use the native-comp feature. I rarely notice any hesitation or slowness. I don't use Elpy. I do use lsp mode.

Since you are doing code development, the first things to go for would be setting up your emacs packaging (installing use-package and melpa (use-package's documentation covers this) so you have more packages to choose from (do be careful to not just pick things willy nilly but research them a bit first)) and then setting up lsp-mode. lsp-mode lets you use LSP servers for the specific programming languages you work with in a somewhat unified fashion. You then need to install and setup the LSP servers for the languages you use, and possibly install language specific Emacs packages as support (note, Emacs has builtin functionality for many).

And first of all, I'm trying to understand, how is it connected to https://github.com/emacs-lsp/lsp-mode? I'm sure, that existed lsp implementations already parse source code. Why TreeSitter?