Introduce an UnenumerableOperations extended attribute

[Ms2ger]

---

Preview | Diff

[domenic]

Thanks for working on this! I have two thoughts:

• Should we provide some sort of guidance on when to use or not use this? It seems quite experimental. The nightmare scenario for me is various spec authors sprinkling this in their specs because they think enumerability is weird, or non-enumerability is more JavaScript like. ("Various spec authors", like myself 5 years ago.)

• Alternately, we could make this part of the "new defaults" inside modules, and not have a separate extended attribute at all.

[annevk]

I agree with @domenic.

If we want this for modules, but are unsure outside of modules, let's limit it to modules.

If we are sure outside of modules, let's make it the default and sprinkle LegacyEnumerableOperations around.

[Ms2ger]

I am not sure this is desirable anywhere, but I updated the PR to make this the behaviour for modules.

[ExE-Boss]

This should also be defined for Stringifiers, Common iterator behavior § forEach and Iterable declarations, like I’m doing in #825.

[littledan]

This PR was motivated by using WebIDL in a way that bridges the gap a bit with JS classes (which might relate to some built-in modules proposals, to facilitate accurate polyfilling) and TC39 conventions. It seems like we're not currently moving in that direction very much at the moment, so leaving this PR hanging (or closing it?) seems like right call.

[domenic]

Let's close this until we have a concrete consumer.