But you have no access to that some_obj.a, unless you export it. After optimizations, it's just a mangled text. Also, I'd think that potentially GCC can even inline such things.

var in clojure is a thing that holds a thing. think of it as an atom that can carry metadata as well as the thing in it.

so no it is not similar in that regard. it is similar if you only care about the thing inside, not the thing holding it

For a cljs file

(ns app.main)

(def foo 42)

(defn -main
  []
  (prn "hello world"))

// Compiled by ClojureScript 1.11.54 {:optimizations :none}
goog.provide('app.main');
goog.require('cljs.core');
app.main.foo = (42);
app.main._main = (function app$main$_main(){
return cljs.core.prn.call(null,"hello world");
});

//# sourceMappingURL=main.js.map

it just isn't possible. look at the code again after :advanced optimizations and you'll understand

pinkfrog needs it only in the REPL though, so optimizations should be irrelevant there, right?

> it just isn’t possible. look at the code again after :advanced optimizations and you’ll understand Looking at the generated “advanced” code, it’s some mangled texted. Continuing the above example, my understanding is the changed things are only the names. app.main is mangled to some random xxxyy names, yet it is still an js object (functioning similar to a ns), and even the attribute foo is also mangled, it still provides indirection.

> pinkfrog needs it only in the REPL though, Yes. The relevant code are only run/triggered from a repl.

sorry I must have missed the part where you explain what you are trying to do

and no app.main is no longer still a JS object with properties. it'll be collapsed and just be something like Xc