Question: what counts as a backing field? A private field that only use in accessors? A private field that's returned from a getter? What about the following?

class C {
  private _f1;
  private _f2;
  get f() {
    return this._f1 + this._f2;
  }
  set f(v) {
    this._f1 = v;
    this._f2 = v;
  }
}

...Or, should we keep the scope very limited and only detect getters that contain a single return this._field? That seems to defeat the purpose of getters though, which is to have additional gating logic.

The initial idea was to only allow the underscore prefix if there was a getter with the exact same name (with or without actually checking that it returned that private field).

In your example, you wouldn't be able to prefix f1 or f2, since there's no corresponding getter with those names, so no conflict.

If that is extended to "any private field that appears to be a backing field", then yes, it would need to check all accessors to ensure that at least one of them referenced the private field.

Different goals I guess. Perhaps there would need to be an option (any backing field, or only backing fields that would otherwise conflict with accessor names).

If it's only "private field with the same name as a getter", then I think it's easily doable. Sounds like a fair heuristic.

Per #8792, we're feature freezing this rule. It's reached the point of complexity where we're having a hard time justifying working on it.

If you'd like to see this option happen, I'd encourage you to add it to a community project such as eslint.style. Cheers! 💙

I also mentioned this just now over in my question to the ESLint Stylistic maintainers about potentially migrating this rule to ESLint Stylistic:

• eslint-plugin: Feature freeze naming and sorting stylistic rules #8792 (comment)