How much does the Wasm backend improve the size of the generated code compared to the JS IR backend?

We haven't integrated dead code elimination to K/Wasm yet. Code size measurements are not very useful until then.

Ah, I see. But it's expected to generate smaller code than Kotlin/JS once DCE is implemented, right?

I would expect higher constant overhead because in K/JS we use built-in library for things like math, strings and arrays. But overhead multiplier for user defined code should be lower because of compact byte-code. To sum it up - smaller size for large projects and bigger size for tiny projects.

Interesting. I noticed that the size of generated code with Kotlin/JS seemed to grow quite quickly as the project grew, so maybe the project doesn't have to be all that big to benefit. Thanks.

These is just my intuition. We'll have to measure and see.

Also can get the benefits of parallelism right