most likely clojure-mode.

One thing is check if clojure-toplevel-inside-comment-form is t. Try setting it to nil and seeing if that helps

I think this could be the “top of defun” feature is breaking with the overload of comment

Oh it's definitely t I use that feature alot

I'll try disabling it and see if the hangs stop.

agree. I do as well

it look like a good mixture of things 🌀 lsp, markdown, clojure-mode

i'm not sure where the markdown is coming from 😕 aside from the fact that the doom markdown package is enabled. not sure why it's running in my clojure-mode though

the root of your stacktrace says lsp-fontlock. you might want to disable it. CIDER already provides sweet font-locking which is not exclusive at all with clojure-lsp

what is font locking exactly?

https://www.gnu.org/software/emacs/manual/html_node/emacs/Font-Lock.html syntax highlighting basically

I suspect this might be related to lsp ui? Maybe the doc popups? Perhaps it grabs the source and puts it inside of a markdown string and renders that?

Not really, it's a lsp-mode core feature, that when hovering in a element, it shows the docs of the symbol, so we font-lock the clojure code

but I do see clojure-mode on the exception, and my guess is there the issue

lsp font-locking just called clojure-mode font locking

but it's wrapped in markdown (as seen in the stacktrace), surely that makes it unparsable to clojure-mode

hum, odd, I think it should not cause any issue as all other LSP langs do the same

clojure-mode should at least ignore I think

ignore what - arbitrary markdown syntax? seems an odd requirement :thinking_face: I'm hypothesizing about how this stuff works, just by judging the stacktrace, but if markdown is passed to clojure-mode, better to strip the markdown delimiters beforehand

Explaining better the feature: • client requests information about the symbol, docs, clojure code example etc via lsp hover request • server response with custom markdown showing all the docs + clojure code as a string. • client(Emacs) just delegate to the major mode to apply font locking on what is inside the clojure markdown, but AFAIK it should not send markdown to clojure-mode

checking out the stacktrace again, the string starts with comment not (comment that's the issue - unbalanced parentheses (either on the buffer itself, or due to some bug in the responsible code)

I think it's expected to have invalid code all the time while developing, shouldn' t clojure-mode handle that better?

depends on the definition of invalid code you should have balanced parentheses 99% of the time which emacs makes really easy although it is to be confirmed that @U70QFSCG2’s buffer actually has unbalanced parentheses

I think we should not expect user has a paren tool or something like that, at least o clojure-lsp, clj-kondo we always need to handle invalid code which is more common that we think

let's keep it simple and determine first the actual issue :)

I use lispyville and lispy to prevent unbalanced parens.

So no.. there aren't unbalanced parens in this case (though it can happen if you paste in text containing unbalanced parens, but that wasn't the case here)

> One thing is check if clojure-toplevel-inside-comment-form is t. Try setting it to nil and seeing if that helps This definitely fixed it.. when nil I don't get any hanging

here is a full stacktrace (the previous was cutoff)

nice. if we can figure out a minimal case we can solve it. don’t think we can use runtime information to know that comment is not clojure.core/comment though

(ns app.bug
  (:refer-clojure :exclude [comment]))


(def comment {:comment/body "I probably shouldn't be using comment"})

I can trigger the bug with just this code

ok. we can make it smarter by ignoring comment forms that don’t have a ( immediately preceding them

I'm not sure how exactly to trigger it.. simply editing things in and round the def seems to trigger it sometimes. Even just using the mouse to place the point somewhere in that form will cause it

That sound reasonable.. until we go and use a comment var as a function xD. But at least in this codebase I don't see myself doing that heh.

yeah. but then if you do that its on you to not use the “comment acts as top level” then i guess

indeed

that sounds fair to me

Should I open a ticket on github?

sure

Done https://github.com/clojure-emacs/clojure-mode/issues/639 Thanks for the assistance folks

> don’t think we can use runtime information to know that comment is not clojure.core/comment though (ns-resolve 'comment)? and friends e.g. ns-interns (maybe you know this well and mean that clojure-mode can't do this without cider?)

this is clojure-mode so we have no connection to a runtime

I think you shouldn't start second REPL. You need to use cider-connect-sibling-cljs command

THANK YOU SO MUCH!!!

matter of fact you should probably figure out the way to use cider-jack-in-clj&cljs that way it would automatically create two REPLs for you. And what's great about it that in .cljc files, if you set the major mode to clojure-mode - it will eval things in clj REPL, if you change it to clojurescript-mode - it will eval things in cljs-repl

I am struggling to do this. I run cider-jack-in-clj&cljs and the Clojure REPL boots ok, but the ClojureScript one says: > error in process filter: ClojureScript is not available. See https://docs.cider.mx/cider/basics/clojurescript for details The docs are not very helpful there. Most likely my configuration is wrong, so I looked up full-stack projects on GitHub and mostly copied https://github.com/schnaq/cljs-re-frame-full-stack but the results are the same. Would you happen to have pointers to help me out here?

here what i do with shadow-cljs: somewhere in your code just require these two lines

[shadow.cljs.devtools.api :as shadow]
   [shadow.cljs.devtools.server :as server]

(server/start!)
(shadow/watch :app)

cider/cider-nrepl                    {:mvn/version "0.28.7"}

Cheers! I will try that.