proposal-top-level-await VS rust-async-bench

For more info about this feature, you can refer to the official proposal repo.

This one is Stage 4 (complete)

But browsers don't support top-level await yet. So we need to put it all inside as async function.

Note that ESM is not “backwards” compatible with CJS: a CJS module cannot require() an ES Module; it is possible to use a dynamic import (await import()), but this is likely not what consumers expect (and, unlike ESM, CJS does not support Top-Level Await).

top-level await feature for ES modules

That said, ES2022 does introduce a top-level await that works in the top level scope, but only in modules, not global scope. So if in a module this could work:

With the latest node version(s) (v14.8+), we should be able to rewrite the above code to something like this. proposal-top-level-await

Now a days, you'll likely mostly use async/await. The few cases where you'll need the Promise API (then()'s etc.) is when you're outside of an async function. With the new top-level await feature coming to modules, you wouldn't even need it outside of async functions in module code which will make its usage even less common.

The await is in top level. The top level await proposal is in stage 3. Currently, the browser support is also not good enough but should work in latest version of the google chrome browser.

There is a proposal for top-level await in JS. I'm guessing this would effectively kinda do the same thing? Or am I wrong there?

Rust's async story is zero-cost in the sense that it is not possible to implement a better asynchronous state machine than the one the compiler will provide for you, but it is explicitly not zero-cost in the sense that there is no overhead to asynchronous execution in comparison to synchronous execution. See experiments like this for examples.