typeof on #private Fields not handled by typeof import()

[magic-akari]

Bug Report

Can you add a new test case?

// a.ts
export class Foo {
  static #p;
  m(param: typeof import("./b").#p) {} 
}
// b.ts
export { Foo as default } from "./a"

The syntax should be valid.

Originally posted by @JLHwung in babel/babel#14454 (review)

Oh I misunderstood typeof import("./foo") before: I thought it was like js import() but it turns out to be actually a dynamic namespace import. So to make type checker happy, we can

// privateNameInTypeQueryImport.ts
export declare class Foo {
    static #p;
    m(p: typeof import("privateNameInTypeQueryImportLib").Foo.#p): any;
}
// privateNameInTypeQueryImportLib.ts
export { Foo } from "./privateNameInTypeQueryImport"

Originally posted by @JLHwung in babel/babel#14454 (comment)

I'm not sure if typeof import('path').#p is valid syntax, but I think typeof import('path').Foo.#p is.

🔎 Search Terms

• typeof on #private Fields
• typeof import

🕗 Version & Regression Information

• This is new feature in TypeScript 4.7 Beta

⏯ Playground Link

Playground link with relevant code

💻 Code

// @filename: main.ts
export class Foo {
  static #m = 1;
  static #f(x: typeof import("./lib").Bar.#m) {}
}

// @filename: lib.ts
export { Foo as Bar } from "./main";

🙁 Actual behavior

!!! error TS1003: Identifier expected.

!!! error TS2694: Namespace '"tests/cases/conformance/classes/members/privateNames/lib".Bar' has no exported member '(Missing)'.

🙂 Expected behavior

• no TS1003 Syntax Error
• TS2694 should report the member name(#m)

[nicolo-ribaudo]

I wrote this example that hopefully makes it clear why intuition suggests that it should be supported:

// test.ts

export class A {
    static b: 1 = 1;
    static #c: 2 = 2;

    method() {
        type A = typeof A.b;
        type B = typeof import("./test").A.b;

        type C = typeof A.#c;
        type D = typeof import("./test").A.#c;
    }
}

type A and type B are equivalent (both are 1), so I would expect also type C and type D to be equivalent. However, the last one is a syntax error.

(ref #47696)

[jakebailey]

The original PR that allowed private identifiers in typeof is begin being reverted in #48959; we may not even allow this syntax based on our design meeting discussions.