Looks like a bug in CLJS compiler, and I couldn't find anything related in Jira: https://ask.clojure.org/index.php/11607/binding-js-variable-to-variable-binds-nil

just ran into this the other day myself, xposting this https://clojurians.slack.com/archives/C03S1L9DN/p1646055931158349?thread_ts=1646040199.234749&cid=C03S1L9DN

@p-himik it's a variant of a known issue around locals, shadowing, and js globals

@p-himik @lilactown are you both trying this at the top-level though?

top-level is maybe somewhat tricky to handle and not a high priority

It's actually @pez that was using it, not me.

@dnolen it was in a let-box in a defn.

what is let-box?

the examples thus far are not so meaningful

and the test at the REPL not so meaningful because of what i said about the top-level

he means a let binding in a defn (I think)

so we need to see something more minimal but complete inside of defn

window is in the global shadow set, so already this is strange

literal code was

(defn start!
  []
  (let [window (js/window.open "" "" "")
        container (.. window -document (createElement "div"))]
    window))

i eval'd this at the REPL, then tried to call it (start!)

@lilactown that's better thanks 🙂 people do top-level let so it wasn't clear

Would a patch be welcome with something like

(core/defmacro js-exponentiate [base exponent]
  (core/list 'js* "~{} ** ~{}" base exponent))

hrm ... does clojure.math provide something here? \cc @quoll ^

@dnolen all I see is a wrapper over Math/pow: https://github.com/quoll/cljs-math/blob/main/src/cljs/math.cljc#L499-L504

@sritchie09 are ** and Math/pow semantically the same other than the boundary across types?

according to https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Operators/Exponentiation, It is equivalent to Math.pow, except it also accepts BigInts as operands.

@sritchie09 it would be preferable for this to be in clojure.math then

I have been looking to expand cljs-math to BigInt, but it's still a work in progress. Sam’s suggestion seems good at face value

I believe @dnolen and @quoll that an older version of the closure compiler didn’t like **, which led to me make this PR to Fraction.js before understanding this fully: https://github.com/infusion/Fraction.js/pull/43#issuecomment-881054662 we should be okay now with the version cljs uses.