Clojurians Log v2
Clojure programming
Channels
Apps
clj-kondo 2026-04-13
borkdude 2026-04-13T09:18:53.941469Z
There's an open PR for this:
(not (nil? x)) -> (some? x)
(when-not (nil? x) ...) -> (when (some? x) ...)
(if-not (nil? x) ...) -> (if (some? x) ...)
Personally I'm not completely convinced that either one is always clearer than the other.
Maybe leave some feedback in the issue: https://github.com/clj-kondo/clj-kondo/issues/2788
Couldn't the same be said for the (not (empty? x)) -> (seq x) one?
borkdude 2026-04-13T12:57:46.378129Z
That one is controversial now since clojure 1.12
@alexmiller those create a binding which might not be needed
if I'm not wrong
Alex Miller (Clojure team) 2026-04-13T13:06:56.750999Z
Ah
> controversial now since clojure 1.12 what's a good documentation of this?
@borkdude to circle back to the OP, I usually prefer the latter forms, i.e. the ones using some? but it can be a style preference thing. Perhaps default to off?
hah
jramosg 2026-04-15T05:38:04.579309Z
related to this, the not-empty? linter does not warn when we use when-not or if-not maybe it should?
> clojure -M:clj-kondo/dev --lint - <<< "(when (not (empty? (map identity [1]))) true)"
<stdin>:1:12: warning: Use (seq x) instead of (not (empty? x)) when x is a seq
linting took 40ms, errors: 0, warnings: 1
> clojure -M:clj-kondo/dev --lint - <<< "(when-not (empty? (map identity [1])) true)"
linting took 33ms, errors: 0, warnings: 0borkdude 2026-06-16T14:58:47.233559Z
I discovered that in hot paths (when-not (nil? ...)) is faster than (when (some? ...)) since some? is a var call
😢 1
👌🏿 1
borkdude 2026-06-16T14:59:01.425559Z
so I'm going to add that warning to the linter docs. it's off by default anyway so it's fine
borkdude 2026-06-16T15:00:11.481209Z
CLJS does optimize some? to != null