Thomas Heller

Not necessary, this is the correct intended core.async version. The rest seems correct too. That leaves anything messing with Classloaders, the JVM in general or maybe clojure itself such as...

2.27.1 included an upgrade in the Closure Compiler, which I guess if the problem here. I'll see if I can figure something out, or maybe downgrade the GCC version again.

One temporary fix is setting `:compiler-options {:output-feature-set :es-unstable}`, which just disables the problematic `RewriteClassMembers` compiler pass. Still trying to figure out if I can maybe tweak this, so I don't...

Hmm react-hook-form works fine for me, even without that `:output-feature-set`. Wether this is suitable for production or not I cannot say. Not sure which kind of features the code is...

I just released version `2.27.2`, which downgrades the Closure Compiler version back to the previous working one. I could not figure out why this fails. Using the npm `google-closure-compiler` to...

I just cut a new `2.28.2` release, using the recently released v20240317 closure-compiler version. It appears that the packages previously failing now compiled. Please give it a try and report...

Any particular reason you are setting `:es8`? I'm really not keen on downgrading again since this version also fixes some other issues.

I'd set the same value for development as in production, which should correspond for the Browser version you are expected to support. It makes zero sense to only set it...

It is fully intended that the properties get renamed. Since it goes through the same `:advanced` optimizations as all CLJS and other Closure code in can get all the benefits...

Hmm never seen this. I guess is that it is `:npm-module` related. `:npm-module` is already rather hacky, and adding the JS support makes it a bit more sketchy I guess....