Clojurians Log v2

https://github.com/clojure-lsp/clojure-lsp/pull/2142

I've reverted to latest instead and that works fine.

https://github.com/clojure-lsp/clojure-lsp-dev-builds/releases/tag/2025.10.21-15.26.33-nightly seems to work for me

any other stacktraces?

oh not really, there is a exception happening

it may be related with the clj-kondo bump, let me check the logs

😨

clojure.lang.ArityException: Wrong number of args (3) passed to: clj-kondo.impl.config/ns-groups
java.util.concurrent.ExecutionException: clojure.lang.ArityException: Wrong number of args (3) passed to: clj-kondo.impl.config/ns-groups

ah, from which file?

humm I recall we are calling that from clojure-lsp

oh really. well, it's not a public function so I warned you ;)

yeah 😅 https://github.com/clojure-lsp/clojure-lsp/blob/c63ba4b6db29e0963e1f8835efa85e2fd21f5e7e/lib/src/clojure_lsp/feature/diagnostics/built_in.clj#L114

this is a backwards compatible code for the unused-public-var linter

I guess you can copy paste the old version.

let me check

but note that I optimized this function so you may want to do the same

hum yeah, for now I will copy the previous one

doesn't look simple to optimize on lsp side

copy the previous one. for the other function from clj-kondo.IMPL.config you can use the public API: clj-kondo.core/merge-configs

;-)

done! testing local

btw it's pretty easy to optimize using memoize, but I restricted memoization per run! invocation so you don't get any memory leaks

makes sense, I may optmize with a simple memoize too

Pushed the fix, should be available soon @seancorfield

I don't know how many times setting-for-ns is called, but if it frequently then optimizing is certainly worth it

looks like it is called almost for every var

in the unused-public-var linter

yes, indeed

this is my version of re-find memoized:

:re-find-memo
                 ;; memoized version of re-find that takes a pattern string and a match string
                 ;; regex creation is cached
                 ;; matches on regex are cached
                 #_{:clj-kondo/ignore [:discouraged-var]}
                 (let [re-pattern-memo (memoize re-pattern)]
                   (memoize (fn [pattern-str file-str]
                              (re-find (re-pattern-memo pattern-str) file-str))))

and then I use this version of re-find in ns-groups. which is why I needed the extra argument

interesting

you could define a global version of this, but I'm fearing that it will eat up memory over time

or you could just factor our the usage of ns-groups from your functions so you only call it once per namespace or so

instead of per var

we could memoize per ns I guess, that's the best we could do without too many changes

Eric already had a message about it in an issue 😄 machinations are underway.

Ah yeah, I was following that issue for some years hehe glad it was done! Will bump it

which issue was it?

On mobile rn, but it was the vertical alignment feature

https://github.com/weavejester/cljfmt/issues/36

It's not everyday you see a 11y issue fixed :) awesome

.https://github.com/clojure-lsp/clojure-lsp/issues/2129#issuecomment-3421981150

Ah yeah, that one too :)

@borkdude any plans for kondo release? Im thinking of cutting a new LSP one in the following days

yeah, I want to work on some issues Thursday

and then doing a release probably

@pez that may be a good addition to calva, I don't recall if calva has workarounds to make that feature work, if so, you can maybe consider relying on cljfmt only now 🙌

I think this means we can get rid of the copy of an old fork of cljfmt we use in Calva for this today.

@pez, @ericdallo you may also be interested in the realign-form function that I've just added to cljfmt 0.15.2. This function will take a rewrite-clj parsed form and aligns the columns of just that structure - so it won't recurse down and align the children's columns as well. This could be potentially useful for a code action of some kind.

As an example, if realign-form is called on this map:

{:x {:aaa 1
     :b 2}
 :yyy 3}

{:x   {:aaa 1
       :b 2}
 :yyy 3}

for sure! would be nice to have 2 code actions for that: • Align vertically this form • Des-align vertically this form (?) the other way around

I see myself wanting the latter one personally, like see a vertically aligned code and want to change to not be aligned like that

Ah, I didn't think of doing it the other way. I guess that's just a case of removing all non-indenting consecutive spaces. I'll add a unalign-form function tonight or tomorrow.

exactly! I mean, this whole vertical alignment feature is really polemic, each person has its own tastes, so I think we should give both options, align or not, convert or not

I can honestly see myself using both. I tend to align into columns only when there won't be too much whitespace (i.e. when the key lengths are similar).

yeah, I can see how convenient it may be for some cases!

:remove-multiple-non-indenting-spaces? does essentially that no?

Yes, but recursively.

I am also not a fan of vertical alignment and I had great success normalizing codebases with that option :)

Glad to hear it! In this case, unalign-form would be a non-recursive option for removing vertical alignment for a single form via a tool like Clojure LSP, Calva, CIDER, etc. The use case is for people who might want to align some forms but not others.

And of course it turns out to be more difficult than I expected...

oh right