Inconsistency with Template Literal Types and Conditional Types

[13OnTheCode]

🔎 Search Terms

"Template Literal Types", "Conditional Types", "${number}"

🕗 Version & Regression Information

5.4.3

⏯ Playground Link

https://www.typescriptlang.org/play?#code/C4TwDgpgBAKhDOwA8MoQB7AgOwCbykQCcBLbAcwD4oBeWNTHfKAJQgGMB7I3FAGigBXbAGtsnAO7ZqAfijAig6AC4oAMwCGAG3gQAUHtCRYABgCMtWAmQADACQBvbIIC2AIwhEAvjcp6A9P5QwQB6MgZG0DAmAEyWcIhI9g5uJORkwD5+gcFQYRHgUSYAzPHWSY7EZORZAUGh4UA

💻 Code

type Test<T extends string> = T extends Record<T, unknown> ? true : false

type T01 = Test<`${number}`> // true

type T02 = Test<`${bigint}`> // false

type T03 = Test<`${string}`> // false

🙁 Actual behavior

inconsistent behavior

🙂 Expected behavior

Additional information about the issue

No response

[jcalz]

I’m struggling to understand what you’re trying to demonstrate. Surely, in general, Test<T> should depend on T, so why would it be “inconsistent” for different inputs to lead to different outputs?

Why would one expect a template literal to be usable as a key to itself? When do you write str[str] and expect that to make sense? Strings have a numeric index signature but not a string index signature or a “bigint index signature” so things look reasonable to me.

I think you should strongly consider explaining the expected/actual behavior instead of assuming it’s obvious.