This PR doesn't have any linked issues. Please open an issue that references this PR. From there we can discuss and prioritise.

What's the part you think we may not have a test of already?

I've looked for one 😅 specifically for a case where a variable with such type would be used as a computed property. All tests seem to only use [Symbol.whatever]. I could miss something but i've grepped the test suite, looked through your PR that has made this possible and looked manually through all current tests in the Symbol "category"

That would be because they're all generic unique symbol tests because my PR essentially just made the SymbolConstructor members into unique symbols (even if not declared as such). And we definitely have unique symbol computed property name tests.

@weswigham I've taken another round throughout the test suite and I can't find a test like this. Knowing the implementation details perhaps this is just the same as:

export class MyObservable<T> {
    constructor(private _val: T) {}

    subscribe(next: (val: T) => void) {
        next(this._val)
    }

    [Symbol.obs]() {
        return this
    }
}

Here, we also have a computed property. However, this has already worked before #42543 and the test case added in this PR wouldn't work the same before that PR.

The closest test that resembles the situation depicted here is this one:

but this file was last touched in 2018 which predates #42543. So since that PR fixed the problem from this added test here, my intuition says that this similar~ test is actually testing something else.

The thing that makes this test here somewhat special is not the computed property itself. It's that this symbol type is first assigned to a variable and the computed property is declared using it.