Clojurians Log v2

Sure

@alexmiller Just a sanity check -- the following doesn't work (it's a reflective call) and we're not trying to fix it with CLJ-2817, right?

(defn ^String primfn-returns-string-1 [^long n]
  (str n))

(defn primfn-returns-string-2 ^String [^long n]
  (str n))

Revisiting this I'm not sure it's an actual issue, just maybe something worth noting. The current patch seems like the natural precedence order.

IIRC I was pointing out that the form

(defn ^Foo f ^Bar [])

Perhaps it's better described as a fix than a regression? I think the actual return type will be Bar at the primitive call site? Before it was incorrectly Foo.

No, I was wrong it doesn't seem to use Foo at all (reflection warning is on Object).

Clojure 1.12.4
user=> (set! *warn-on-reflection* true)
true
user=> (defn ^Integer foo ^String [^long n] (str n))
#'user/foo
user=> #(.length (foo 0))
Reflection warning, NO_SOURCE_PATH:1:2 - reference to field length on java.lang.Object can't be resolved.
#object[user$eval139$fn__140 0x997d532 "user$eval139$fn__140@997d532"]

Okay, that makes me feel better about not being able to create a test that differentiated the two metadata sites 🙂

I'd love to see this merged. I'm ready to tune the patch if it helps to fast-track it. But from what I gather, it is currently fine as it is.

@alexyakushev Did you vote on the Asks?

https://ask.clojure.org/index.php/13501/primitive-type-hints-arguments-turns-the-return-value-object in question