[no-shadow] false positive for generics, callback argument names and more - considered as shadowing since v4.0.0

[bengry]

• I have tried restarting my IDE and the issue persists.
• I have updated to the latest version of the packages.
• I have read the FAQ and my problem is not listed.

Repro

Two issues regarding the same rule no-shadow arose after upgrading to v4.0.2 (and reproduces onn v4.0.2-alpha.2 for the following configuration. v3.10.1 was fine for the same code.

{
  "rules": {
    "no-shadow": "error"
  }
}

I opened 1 issue since it's all on the same rule, but LMK if I should open separate issue for the separate cases:

1. Callback arguments and parameters:

function FilterListItem<T extends string>({ option, onChange }: {
	option: T;
	onChange: (option: T) => void;
}) { ... }

This reports the following error:

error 'option' is already declared in the upper scope no-shadow

2. Generics in static methods

export class Selector<K extends string> {
  public static someStaticMethod<K extends string>(): void;
}

In reality this is a React component, and the static method is getDerivedStateFromProps, but it's the same idea.

This reports the following error:

error 'K' is already declared in the upper scope no-shadow

3. Mapped types

export type DeepReadonly<T> = T extends Primitive
  ? T
  : T extends Array<infer U>
  ? ReadonlyArray<DeepReadonly<U>>
  : T extends Map<infer K, infer V>
  ? ReadonlyMap<DeepReadonly<K>, DeepReadonly<V>>
  : T extends Set<infer M>
  ? ReadonlySet<DeepReadonly<M>>
  : { readonly [K in keyof T]: DeepReadonly<T[K]> };

This reports the following error:

error 'K' is already declared in the upper scope no-shadow

Also related:

export type DeepMutable<T> = T extends Primitive
  ? T
  : T extends ReadonlyArray<infer U>
  ? Array<DeepMutable<U>>
  : T extends ReadonlyMap<infer K, infer V>
  ? Map<DeepMutable<K>, DeepMutable<V>>
  : T extends ReadonlySet<infer U>
  ? Set<DeepMutable<U>>
  : { -readonly [K in keyof T]: DeepMutable<T[K]> };

reports:

317:33 error 'U' is already declared in the upper scope no-shadow
319:18 error 'K' is already declared in the upper scope no-shadow

Expected Result
No ESLint errors for these examples.

Actual Result

Errors are reported, as shown above.
Additional Info

Versions

package	version
@typescript-eslint/eslint-plugin	4.0.1
@typescript-eslint/parser	4.0.1
TypeScript	4.0.2
ESLint	7.8.1
node	12.18.3

[bradzacher]

(1) is already fixed in master.

As per the release notes, you need to use @typescript-eslint/no-shadow, not the base rule.

(2) is correctly reporting.

You have shadowed the generic definition on the class with the generic definition on the function.

It doesn't matter that its a static method, both examples define generics of the same name. The method is a child of the class, hence it shadows.

Example of when this can be confusing

class Foo<T> {
  static method<T>(): T {} // which T is this referring to?
}

const result = Foo<string>.method<number>();
// what's the type of result?

(3)(a) is correctly reporting.

infer K defines a type named K.
K in keyof T defines a type named K

(3)(b) is also correctly reporting.

You have two instances of infer U.
And the same issue as before for K.

[bengry]

@bradzacher Sorry for the long delay, haven't had the time to look into it again until now.
Re 2. If you remove the generic from the static method you get a TS2302 error ("Static members cannot reference class type parameters"), so I think that this one is incorrect. Sure, I can rename the static method's K to some other name and have the warning gone, but since I can't reference the class' generic K anyway - might as well use the same name and have ESLint not report it.