Hi!

Yes, of course, you can use clj-kondo from the REPL:

$ clj
Clojure 1.11.0
user=> (require '[clj-kondo.main])
nil
user=> (require '[clj-kondo.core :as clj-kondo])
nil
user=> (:findings (with-in-str "(+ :foo)" (clj-kondo/run! {:lint ["-"]})))
[{:filename "<stdin>", :row 1, :col 4, :end-row 1, :end-col 8, :type :type-mismatch, :message "Expected: number, received: keyword.", :level :warning}]

Thanks for looking into the issue. Let me know if you need more info.

Oh, let me also try the jack-in in Calva

Works fine over here. I chose jack-in, deps.edn, alias clj-kondo/dev

Thanks! This is among my first dive into a real-world clojure project, so I appreciate the guidance

[matcher-combinators.test :refer [match?]] is how I solve this

you might also have to add cljs.test in the config

adding cljs.test fixed it! thank you

I can't reproduce this locally. Can you make a standalone file including extension and upload it here? Back in an hour.

Ugh, I created a test file and get the warning in one project, but not in another, even with the config.edn moved out of the way. Looks like I need to work a little more on reproducing this.

Here is the example as a file. Sometimes I get warnings, sometimes I don't, still trying to figure out what triggers the different results. (cache, configuration, copied configurations, etc.)

Usually I get the following output when checking the file above, but sometimes I don't. Haven't found a clear pattern yet.

$ clj-kondo --version
clj-kondo v2022.10.05

$ clj-kondo --lint inline_spec.clj
inline_spec.clj:4:10: warning: inline def
inline_spec.clj:7:28: warning: inline def
linting took 12ms, errors: 0, warnings: 2

Nevermind, I found the following nonsense in my global config.edn file.

{:lint-as {clojure.spec.alpha/keys clojure.core/def}}

Ah, glad it's solved now :)

Here is the definition for it:

[{:end-row 30
     :meta nil
     :name-end-col 11
     :name-end-row 17
     :name-row 17
     :ns com.fulcrologic.fulcro.dom
     :name a
     :defined-by cljs.core/declare
     :filename "fulcrologic/fulcro/src/main/com/fulcrologic/fulcro/dom.cljs"
     :col 1
     :name-col 10
     :end-col 45
     :row 17}

Whereas the generated docstring for all of them is like this:

*com.fulcrologic.fulcro.dom/div*
  [& args]
  Returns a React DOM element. Can be invoked in several ways

These two are made equivalent at compile time
(div "hello")
(div nil "hello")

These two are made equivalent at compile time
(div {:onClick f} "hello")
(div #js {:onClick f} "hello")

There is also a shorthand for CSS id and class names
(div :#the-id.klass.other-klass "hello")
(div :#the-id.klass.other-klass {:onClick f} "hello"))

The :var-usages info also does not include :arity:

{:end-row 6
     :name-end-col 51
     :name-end-row 6
     :name-row 6
     :name div
     :filename "ui/reports/function_profile.cljs"
     :from ui.reports.function-profile
     :col 48
     :name-col 48
     :end-col 51
     :refer true
     :row 6
     :to com.fulcrologic.fulcro.dom}

(div :.sixteen.wide.column.segment "test")

To make this more tooling friendly, one could choose to generate the code to disk instead of only via a macro at compile time

Or write a hook which expands this call into stuff that clj-kondo understands better

Is there a way to write a single hook that would cover all of those generated items? They essentially expand to the same kind of form and all have the same (almost) docstring.

a single hook yes, but you would need to have a mapping for all of those vars in the config currently

so {foo/a hook foo/b hook foo/c hook} etc

That would only help with the :var-usage data and not the :var-definition, right?

Haven't written any hooks myself yet..

that would help with var-definition since you can expand this into a hook which generates the right var definition

Alright, thank you. I will do it for these Fulcro items. Are you otherwise interested in receiving reports about such analysis peculiarities? I'm running clj-kondo analysis on many libraries and trying to fit the results into a normalized SQL schema. In most cases things work just as they should but there are quite a number of mismatches across the ecosystem. By mismatches I mean cases where a var usage doesn't match up to some expected var definition foreign key.

Yeah, sure, anything that can improve the quality of the analysis is surely welcome

And I appreciate your work on this so far

Happy to help! I'm thinking if I create a mismatch report it could help identify the libraries that need the most help or have the biggest ecosystem impact.