Kevin Gibbons

> Promise.any reaches stage 3 With a caveat for first changing `.errors` to a prototype accessor

To be clear, public/private instance were already at stage 3. This is public/private static to stage 3.

~~Note that you may want to hold off work until after the July TC39 meeting, in case this gets shot down.~~ J/k you should do it now :)

A bit of detail: this happens for any non-writable attribute on the parent, not just for frozen objects; it's specified in step 3.a of [`OrdinarySetWithOwnDescriptor`](https://tc39.github.io/ecma262/#sec-ordinarysetwithowndescriptor) (which is called by `OrdinarySet`...

Apart from this, there's also the `argumentsObjectNeeded` stuff, which is unobservable (except that you do need to avoid creating the binding for arrow functions, of course). Similarly, the condition containing...

> while I think that the `then` method of a thenable itself is relevant too The only time this would matter is if you had a `thenable` whose behavior was...

Yes. My point is that as a _producer_ of thenables you should not design on the assumption that `then` is only called asychronously, because this is not true; rather, your...

> fortunately you don't have to care about that My claim is that as a consumer you don't have to care about it either way. This PR would not change...

It does not make changes relevant to producing. Producers already have to be aware that `then` may be invoked synchronously or asynchronously. This PR does not change that.

It is simply a fact that `then` may be invoked synchronously or asynchronously. Yes, it's possible to write code which would break if if `then` were invoked synchronously but where...