string literal type defined via a mapped type is sometimes erroneously interpreted as simply string

[seansfkelley]

A "direct" string literal type defined using type Foo = "A"; behaves properly in more places than an apparently-equivalent "indirect" string literal type defined using mapped types.

TypeScript Version: 2.3.4

Code

I discovered this issue while using https://github.com./dphilipson/typescript-string-enums, and as such, have inlined the relevant part of the library into the top of the example.

// Inlined, stripped-down version of the relevant part of typescript-string-enums.

function Enum<V extends string>(...values: V[]): { [K in V]: K };
function Enum<
    T extends { [_: string]: V },
    V extends string
>(definition: T): T;
function Enum(...values: any[]): object {
   // ... elided for brevity ...
   return {} as any;
}

type Enum<T extends object> = T[keyof T];



// Version 1: Using a "direct" string literal type.

// type MyEnum = "ENUM";



// Version 2: Using an "indirect" string literal type.

const MyEnum = Enum("ENUM",);
type MyEnum = Enum<typeof MyEnum>;



// The offending code sample.

interface MyInterfaceWithEnum {
    foo: MyEnum;
}

function parseToEnum(s: string): MyEnum {
    return null as any;
}

function failingFunctionWithMap(s: string): MyInterfaceWithEnum[] {
    // This block fails to compile when using an "indirect" string literal type, claiming that
    //     Type 'string' is not assignable to type '"ENUM"'.
    // This implicit change in type does not happen with the "direct" string literal type.
    return [""].map((i) => ({
        // Note that VSCode correctly reports the output of `parseToEnum` here for both enums.
        foo: parseToEnum(i),
    }));
}

function successfulFunctionWithoutMap(s: string): MyInterfaceWithEnum {
    // This block always compiles, using either of the "direct" or "indirect" string literal types.
    return {
        foo: parseToEnum(s),
    };
}

Expected behavior:

Both definitions of MyEnum are equal and interchangeable.

Actual behavior:

The "indirect" type erroneously fails to compile in one of the two functions.

[DanielRosenwasser]

I believe that this is a duplicate of #11152.

[seansfkelley]

Certainly it's very closely related, but I think it's possibly different, since the type inference in my example does work as expected simply by changing the way the type is defined. It seems plausible that a fix to #11152 would also fix this, but I don't know enough about the compiler to state that for sure nor to definitively declare this as a duplicate.

[mhegazy]

Automatically closing this issue for housekeeping purposes. The issue labels indicate that it is unactionable at the moment or has already been addressed.