Clojurians Log v2
Clojure programming
Channels
Apps
clj-kondo 2026-09-17
mpenet 2026-09-17T13:34:29.374969Z
hey! any plans to implement new linters on top of :excess & :missing? like levenshtein distance hints vs known params : "you wrote :for-update but did you mean :for-update?" kind of warnings
🙏 1
borkdude 2026-09-17T13:57:28.694759Z
clj-kondo already checks keys that are missing, what is still missing?
mpenet 2026-09-17T13:58:45.877189Z
Check for typos on calls to fns taking maps with destructuring (:keys)
borkdude 2026-09-17T14:01:33.861759Z
if a key is not required, how can you be sure it's a typo
anthony-galea 2026-09-17T15:15:39.618399Z
@borkdude we can’t be sure, but perhaps we could suggest when we think it’s likely: ex:
(defn foo [{:keys [for-update?]}])
(foo {:for-update true})borkdude 2026-09-17T15:18:52.628289Z
but yeah we could make a linter for this that is turned off by default
borkdude 2026-09-17T15:19:11.527309Z
not sure what the impact on performance is (probably ok)
borkdude 2026-09-17T15:19:41.580729Z
it would probably would have to be integrated with the type system since that records all the accumulated keys
mpenet 2026-09-17T13:35:18.738169Z
well I guess it's not necessarily bound to excess & missing
borkdude 2026-09-17T15:48:50.950339Z
I'm proposing the following change: if a namespace is used in a hook (macro-expansion), clj-kondo will consider it as already being required. This allows you to use macros that use other namespaces without requiring those namespaces (as is often the case in clj-kondo). I think the false positives outweigh the false negatives here.
that's probably fine, yeah
what are the false negatives?
borkdude 2026-09-17T17:52:09.942459Z
false negative could be when e.g. a clojurescript JVM macro expands into a namespace usage that isn't required at runtime. but very rare I'd say. and no regressions found
👍 1
in that case, i'd say you're good to go