satisfies alters type inference of some (generically-typed?) variable assignments

[stwlam]

🔎 Search Terms

satisfies narrows generic assignment

🕗 Version & Regression Information

This is the behavior in every version I tried (from 4.9.5), and I reviewed the FAQ for entries about Generics

⏯ Playground Link

https://www.typescriptlang.org/play?ts=5.3.2#code/C4TwDgpgBAChxQLxQEQGMCGwVQD6oBMB7AcxQG4AoNIgOwGcFJh6koBtdLFAGkNJQBdKPSwBLegDMxEVnGDtBVZvQB0Y2mgA2AVwKyAFCgBGYgE4EUASigB6W1FpEERANZA

💻 Code

type Pet = "cat" | "dog";
const pets = ["cat", "dog"] satisfies Pet[];
pets.includes("bird") // not ok

🙁 Actual behavior

pets is of type Pet[]

🙂 Expected behavior

pets is of type string[]

Additional information about the issue

No response

[fatcerberus]

Yes, satisfies induces contextual typing, so it can sometimes affect the type of an array or object literal. If it didn’t, then the above would fail because string[] isn’t assignable to Pet[].

[stwlam]

I suppose the announcement reads a little off, then: perhaps when it's documented it could be clarified further?

The new satisfies operator lets us validate that the type of an expression matches some type, without changing the resulting type of that expression.

[fatcerberus]

I agree that should be clarified, but to be fair, technically if pets is allowed to be widened to string[] then it no longer satisfies Pet[] right? 😉

[stwlam]

I assumed it would confirm that the value is assignable to the satisfies type, narrowing as necessary, but then returning the actual type assignment to normal inferencing procedure.

[jasikpark]

I think I would consider the strict array.includes behavior more the issue, no? a la https://github.com/total-typescript/ts-reset

[jcalz]

Duplicate of #55189… the documentation is indeed misleading here