extends operator with literal types

[dpohvar]

Bug Report

🔎 Search Terms

🕗 Version & Regression Information

• This is wrong behaivor of extends operator
• This changed between versions 3.3.3 and 3.5.1
• Most likely this is the reason: Relate source types covered by a target discriminated union #30779

⏯ Playground Link

https://www.typescriptlang.org/play?ts=5.1.0-beta#code/C4TwDgpgBAglC8UDeAPAXAZ2AJwJYDsBzAXygB9l18BXAWwCMJtSLU0oMQGB7AG2IBQA0JCgAhBJUw4CJclJoMmLKVHrc+EAIb5BUfQcNHjJ0wYD05qAD0ovXMCZbeUEdAAWTCMPDQAwpJsWHhEKmyKjMxCblAAGoHowbLyEcoCllAABvEQKI74ACYY4plq1MBQ+NwVcNzYUH5CPqKxAPpwiDl5EIXFcAD8UABEXflFsENQ7COV1VC5Y31DANzpVvrW-dG+ca0SnfPdveJQgzMLPeNik9PxVRUXx9erGRtbzdBtAQeP4wFno0uxT8N2Gdzmv2BKzWBk221EMFa8UQcEhcVOw1RR3GsVBQzg90Oizi0NeNneMTESMkEjR8TOtOxxVxU2GEkJdNJ63J8P81MQATpGKGgqZJNZItmDzFuJe3LhQA

💻 Code

type A = {x:string} | {x:number} | {x: symbol}

type B = {x:string} | {x:number} | {x: boolean}
 // ^ literal type here
type C = {x:string} | {x:number}

type X = {x:string | number}
// `X extends B` but not A or C


type X_A = X extends A ? "X extends A" : "X not extends A";
// ^? X not extends A

type X_B = X extends B ? "X extends B" : "X not extends B";
// ^? X extends B

type X_C = X extends C ? "X extends C" : "X not extends C";
// ^? X not extends C

the bug is reproduced with any literal type instead of boolean

🙁 Actual behavior

X not extends A
X extends B;
X not extends C

🙂 Expected behavior

X not extends A
X not extends B
X not extends C

[jcalz]

#30779 implemented this behavior intentionally, and it is documented in the TS3.5 release notes. Unions with discriminant-worthy properties (involving literals) are related differently, on purpose. That results in:

type A = { x: string } | { x: number } | { x: symbol }
type B = { x: string } | { x: number } | { x: boolean }
type C = { x: string } | { x: number }

type X = { x: string | number }

const x: X = { x: Math.random() < 0.5 ? "abc" : 123};
const a: A = x; // error
const b: B = x; // okay
const c: C = x; // error

And so extends is behaving the way assignability behaves. Maybe it's inconsistent, but it's not a bug.

[github-actions]

This issue has been marked as 'Not a Defect' and has seen no recent activity. It has been automatically closed for house-keeping purposes.