It works by yielding. Whenever more data arrives, the wasm starts running to process it, then it is suspended and control is given back to the browser until another piece of data arrives. Thus, the browser can process things such as mouse events while the wasm is suspended.
I haven't looked closely into the details, but I am pretty sure that the WASM modules provide a poll function, and that JS will call the poll function from the appropriate JS event handlers when something happens. It yields by simply returning from this function.
download url1: and callback foo1
download url2: and callback foo2
and the downloads/GET requests happen in concurrent, and we don't know whether foo1 or foo2 is called first
===
How do we express this program in async rust? does wasm have multiple entry points, one for foo1, one for foo2; or is there only one entry point but it somehow distinguishes if it's going foo1/url1 or foo2/url2 ?
You have in JS Promise
In rust pub async fn generates a function that returns a Future structure with poll()
The #[wasm_bindgen] creates a JS function Promise that is like calling tokio::spawn(run())
It gets called here.
Downloading does not happen in WASM it is done using Fetch
To do concurrent downloads from single function call either;
Set code to be call in then after window.fetch_with_request(&request)
or join or select Futures that come JsFuture::from(window.fetch_with_request(&request))