I don't like it myself, but seg.mnt/ve/yCom|pLex.stu-ff (per https://github.com/clojure-emacs/clojure-mode/pull/475/files) appears to be a valid name, in the same way that clojure.core has (defn / []) for division

keywords as well, or only symbols?

$ clj
Clojure 1.11.1
user=> :a/a/a
:a/a/a

fair point, should've tried that myself

A GH issue would be most welcome The only input we have so far is https://github.com/clojure-emacs/cider/issues/3356#issuecomment-1601174777

https://clojurians.slack.com/archives/C0617A8PQ/p1691916711262449?thread_ts=1691749617.883889&cid=C0617A8PQ

Yeah, that's what I'm thinking of in term of a compatibility layer. You are spot on there.

I wonder how it would work if clojure-ts-mode just defined functions with the exact same name in clojure-mode, like clojure-find-ns for instance, and then at the end of the file called (provide 'clojure-mode) . As long as clojure-ts-mode is loaded before cider this would work. Whatever the solution, it shouldn't be noticed by users IMO

(defvar clojure-ts--find-ns-query
  (treesit-query-compile
   'clojure
   '(((source (list_lit
               :anchor (sym_lit name: (sym_name) @ns)
               :anchor (sym_lit name: (sym_name) @ns-name)))
      (:equal @ns "ns")))))

(defun clojure-ts-find-ns ()
  "Return the name of the current namespace."
  (let ((nodes (treesit-query-capture 'clojure clojure-ts--find-ns-query)))
    (treesit-node-text (cdr (assoc 'ns-name nodes)))))

> I wonder how it would work if clojure-ts-mode just defined functions with the exact same name in clojure-mode Not sure if I would be a fan myself, since older packages would still depend on clojure-mode. So order would be determined by the package dependency tree? Wouldn't sound fun to troubleshoot / support 🙃

Haha, fair enough. Just a fleeting thought.

I'd prefer to make the compatibility more explicit myself - like the checks proposed above for CIDER to just wrap whatever functionality it needs from clojure-mode in functions that try both clojure-mode and clojure-ts-mode. (or something along those lines) Perhaps there can be even a config option in CIDER for which major mode to prefer.

Shouldn't be very hard to do - I guess the only tricky part would be some REPL-powered font-locking and indentation when you're using clojure-ts-mode, but we can simply disable this in the beginning.

> I guess the only tricky part would be some REPL-powered font-locking and indentation when you're using clojure-ts-mode, but we can simply disable this in the beginning That is going to be tricky

It's not essential, so I assume most people will be able live without it for a while. I'm not sure we'll be able to update the TS grammar's on the fly the way we've been doing it for clojure-mode.

The grammar we cannot update on the fly since it is statically compiled

But some of the rules we may be able to, although many of the tree-sitter facilities are setup such that they are captured when the mode is initialized.

We'll figure something out. As I said that's not an essential feature, so I don't see it as some roadblock.

Is it intended that cider-jack-in-clj simply freezes emacs while it loads? I am using Emacs 28.2 and just ran M-x straight-pull-package-and-deps and M-x straight-rebuild-package to try out the latest CIDER

the previous behavior I would see the leiningen command invoked in my message buffer and wait — it was async

these changes also seem to have broken a few leiningen options (e.g. repl options like :init-ns doesnt work anymore)

It can appear to freeze since it's doing stuff. Can take a good minute on a large project

> these changes also seem to have broken a few leiningen options (e.g. repl options like :init-ns doesnt work anymore) Interesting. I honor :init but hadn't thought of :init-ns

:global-vars doesnt work either

in short I have a project with the following set

:repl-options {:init-ns foo.bar}
:global-vars {*warn-on-reflection* true}

I'll see what I can do about that one. Maybe it's easy. Note that Clojure CLI doesn't offer that, so it's in a grey zone

> Is it intended that cider-jack-in-clj simply freezes emacs while it loads? I infer from your other feedback that the command eventually succeeded?

Yes, it succeeded. It just froze all of Emacs until the REPL appeared. The previous behavior in CIDER was spawning lein and I was able to do other things while waiting

Thanks. I'll consider it. It's certainly a nuisance. With Enrich's caching (which is partly shared across projects - much like ~/.m2), as you use it more, load times will decrease. With a bit of luck I can simply use whatever async facility made this work for traditional jack-in.

Is it possible to downgrade CIDER in the meantime? I assume this would depend on the package manager on uses in Emacs

cider-enrich-classpath defaults to nil - you got it changed at some point

yeah, I have it set to t by default since I find the functionality useful, though the current workflow is disrupting a few of my lein projects

in particular, im noticing the dev profile is not being applied (though this is set by myself in a custom lein command in cider)

Does it really take more than a few seconds after the first time though? Profiles should be applied, I've used that myself every day let's debug it

i.e. I have cider-lein-parameters set to "with-profile +dev repl :headless :host localhost", but one of my larger projects is no longer applying some stuff in the dev profile. itll be tough to make a repro but I can try soon

I havent tested startup time when restarting cider, but I can try that right now

my cider-lein-parameters are:

"with-profile \"+async-checking,+test,+dev,+emacs-backend,+emacs-backend-init,+yourkit,+g1gc,+parallel-reload,+cider-nrepl-latest\" repl :headless :host localhost"

I see. I just attempted restarting CIDER, though the REPL only clears itself. However, if I quit the REPL then run M-x cider-jack-in-clj, Emacs freezes for about 10 seconds, but this is way faster than the first startup 🙂

Please paste the output of M-: (cider--extract-lein-profiles cider-lein-parameters) ? The other thing to keep in mind is, as mentioned, :init-ns and :global-vars are disregarded. Which is not a profiles bug, it's something else

("with-profile +dev " "repl :headless :host localhost")

It looks like the profiles do work. I just did a search in the java invocation and indeed I see the extra source paths provided by dev. Sorry for the distraction

No issue! > Emacs freezes for about 10 seconds, but this is way faster than the first startup Pretty good news! > I just attempted restarting CIDER, though the REPL only clears itself. Ah, I'm not a sesman-restart user myself so it escaped my mind. It's an important thing to support. Thanks for the round of feedback! I've gotten awesome value from it.

To you and anyone else reading - I have to take a pause from CIDER hacking till Friday, then I will happily push the next iteration.