(ns foo
  (:require
   [clojure.string :refer [split-lines]])) ;; gives unused-namespace + unured-reffered-var

#?(:clj
   (split-lines ""))

If I move the require to a #?(:clj ) it works, but that way should work as well, right?

clojure-lsp removes the require from ns when the require is unused, this kind of issue could be really bad for clojure-lsp clean-ns feature 😕

I see the problem, but I think the analysis info provides enough info for the clean-ns to not remove this: e.g. one usage is enough to not clean it

from the clj-kondo perspective: the linting happens twice, once for the clj branches and once for the cljs branches

yeah, but it seems a workaround right? fix the root issue seems the right choice?

even if we fix on clojure-lsp, it's still a warning on clj-kondo that is a false-positive 😕

> from the clj-kondo perspective: the linting happens twice, once for the clj branches and once for the cljs branches Is that easy to fix?

Should I open an issue?

no, this is not an issue, it is just how it works and how it has always works: it's intended

😮 even if it's a valid used code, is it right report as unused?

it is unused in cljs. there's an issue to add to the linting "cljs" so it's more clear

oh, got it

https://github.com/clj-kondo/clj-kondo/issues/1196

I see

I would prefer not to check the analysis for performance reasons as ATM we are only relying on findings

We would need to rely on analysis only because of this specific case

we can add some more info to the findings perhaps: a lang key

and if there are not two the same findings for both languages, you should not remove?

hum, maybe

or we can enhance the unused-* linters to behave the way you expected them

with a setting

yeah, that would be the perfect from clojure-lsp view, but that other idea of enhance the findings would work as well I think

with the enhanced findings you would not be depending on what the user would configure, so that seems more stable right?

Exactly

another thing to look out for is when we will change the message to include the language (clj or cljs) then LSP should only look at other fields like :lang and :ns and :name perhaps

I'm not sure if we already include those

are you depending on the exact format of the message in some cases? I hope not?

No, only the code

invalid-arity for example

https://github.com/clojure-lsp/clojure-lsp/blob/master/src/clojure_lsp/feature/code_actions.clj#L39-L41

cool

Hey, what's the easiest way to setup neovim with clojure-lsp? neovim now has the lsp client built in right?

I want @U051N6TTC to try out clojure-lsp ;)

Yes, It has it built-in, but I don't know anything from nvim side besides that :/

Perhaps @U11EL3P9U knows?

He needs to install clojure-lsp, first, this is what I know hahaha

quoll = she

I think he uses CoC

Maybe @UMMMKKADU knows, he uses nvim

She can check this: https://clojure-lsp.github.io/clojure-lsp/clients/#vim

I'm using a super fancy nvim/fennel setup, but as @UKFSJSM38 said Nvim 0.5 has a built-in LSP and you just need a plugin to setup which is this one: https://github.com/neovim/nvim-lspconfig

This in company of https://github.com/hrsh7th/nvim-compe for auto completion and https://github.com/nvim-telescope/telescope.nvim#neovim-lsp-pickers for fancy menus and pickers is all you need

Thanks, we really need a tutorial mentioning all of that :p

coc-nvim bundles all of this in one, but is written in nodejs, people is jumping out

Hahaha I will, I just migrated for the native stuff, but you right.