@bbatsov, could you please take a look?

Sure. At a glance the PR looks OK, but I'll take a deeper look tomorrow, as the changes are quite extensive.

You can close#98 and perhaps add some notes to the README that we're using a modified version of the upstream grammar and our reasoning for doing this. Probably some of this should also go in some form to the design document.

perhaps add some notes to the README that we're using a modified version of the upstream grammar and our reasoning for doing this.

This is already done, but let me know if I should extend it further.

Probably some of this should also go in some form to the design document.

Yeah... I'm still struggling to even start with that :)

This is already done, but let me know if I should extend it further.

For the README that's fine, but for the design doc I'd probably mention the summary from the end of the discussion about the creation of the experimental grammar. Overall the PR seems to be in a good shape to me, but I left small remarks here and there.

I fixed all the mentioned issues and updated the design documentation.

Not sure if it's pilot error on my end, but my Emacs 30.1 doesn't seem to have the functiontreesit-query-valid-p mentioned in this bit:

(defunclojure-ts--clojure-grammar-outdated-p ()"Return TRUE if currently installed grammar is outdated.This function checks if`clojure-ts-mode' is compatible with thecurrently installed grammar.  The simplest way to do this is to validatea query that is valid in a previous grammar version but invalid in therequired version."  (treesit-query-valid-p'clojure '((sym_lit (meta_lit)))))

I think it got added to Emacs inthis commit.

It looks like it got added in 2025-01 (before Emacs-30.1 was released), but may be it didn't make it in?

I don't see it viagit log locally in a checkout of theemacs-30.1 tag.

Fairly short implementation though...

(defuntreesit-query-valid-p (languagequery)"Return non-nil if QUERY is valid in LANGUAGE, nil otherwise."  (ignore-errors    (treesit-query-compile language queryt)t))

Hmm, seems I don't have it as well. It's interesting that lint job didn't catch this...

Wow, that's surprising. I should have tested it on Emacs 30. Should we just copy the implementation directly toclojure-ts-mode.el?

@rrudakov I guess that would be the simplest thing for now. Hopefully it will be included in 30.2 - I heard it's right around the corner, but I'm not sure what's going to land there.

@rrudakov I guess that would be the simplest thing for now. Hopefully it will be included in 30.2 - I heard it's right around the corner, but I'm not sure what's going to land there.

I'll create a PR shortly.