Re: [whatwg/webidl] Add `async iterable<T>` type to WebIDL (PR #1397)

So during an experiment we ran for `new Response(async iterable<Uint8Array>)`, we discovered [another fun edge case](https://github.com/whatwg/fetch/issues/1291#issuecomment-2309773461) to think about: when having a union of `string | async iterable<Uint8Array>`, what should a boxed string resolve as? In the current spec text it resolves as a `async iterable<DOMString>`, which would cause an error during iteration. This happens because the boxed string is both an object, and has a `[Symbol.iterator]` method.

We have two options here:
1. handle this case specially in WebIDL by banning boxed string as a value for `async iterable`, like we do with primitives (kinda weird, but means this doesn't become a weird edge case only in fetch and is instead uniform across the platform)
2. handle this as a weird edge case in `fetch`, and in the future in any other `string | async iterator` WebIDL unions.

The big question is whether, assuming `ReadableStream.from("asd")` throws,  `ReadableStream.from(new String("asd"))` should throw, or create a stream of string chunks.

Any thoughts? 

-- 
Reply to this email directly or view it on GitHub:
https://github.com/whatwg/webidl/pull/1397#issuecomment-2309798759
You are receiving this because you are subscribed to this thread.

Message ID: <whatwg/webidl/pull/1397/c2309798759@github.com>