Fixed mapped types not being considered as homomorphic with substitution constraints

[Mateusz Burzyński (Andarist)]

ports microsoft/typescript-go#4596
fixes #63132

What

This PR fixes the issue where conditionals like this keyof T extends X ? Truthy : Falsy would make { [K in keyof T: T[K] } not being recognized as homomorphic within the Truthy branch of the conditional type.

Extra

This PR comes with a type-fest break. I already looked through it... so I'll just quote my original answer here:

A distributive homomorphic type requires to be written in this form (roughly):

type M<T> = {
  [K in keyof T]: T[K]
}

A type-fest-based repro case can be seen here. As we can see, that exact form ([K in keyof T]: ...) is used there. Therefore, I think the expectation should be that this would be treated as a distributive homomorphic type.

But because of the issue fixed by this PR, that mapped type was not recognized as homomorphic and distributive. That happened because keyof T in [K in keyof T] became (invisibly!) a substitution type internally. With the fix here, it gets recognized properly~ so it started to distribute and thus it broke type-fest's implementation as it was accidentally leveraging the incorrect behavior.

It's common to change the code a little bit to enable/disable distributivity of conditional types and I think the same principle kinda applied to mapped types, given the distributivity (or the lack of it) is also an implied trait of any mapped type.

I just pushed out a test documenting this and showcasing 2 possible workarounds for the type-fest's case, see this commit. If this PR lands then I will go to type-fest and PR a fix for them.

[typescript-automation]

This PR doesn't have any linked issues. Please open an issue that references this PR. From there we can discuss and prioritise.

[Copilot]

Pull request overview

Fixes homomorphic mapped-type detection when keyof T is wrapped in a substitution constraint.

Changes:

• Unwraps substitution constraints before detecting homomorphic mapped types.
• Adds regression and distributivity tests with baselines.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tsc/internal/checker/checker.go	Corrects homomorphism detection.
tsc/testdata/tests/cases/compiler/mappedTypeNotMistakenlyHomomorphic2.ts	Tests issue #63132.
tsc/testdata/tests/cases/compiler/mappedTypeHomomorphismWithPriorKeyofTChecks1.ts	Tests distributive and non-distributive forms.
tsc/testdata/baselines/reference/compiler/mappedTypeNotMistakenlyHomomorphic2.types	Records inferred types.
tsc/testdata/baselines/reference/compiler/mappedTypeNotMistakenlyHomomorphic2.symbols	Records symbols.
tsc/testdata/baselines/reference/compiler/mappedTypeHomomorphismWithPriorKeyofTChecks1.types	Records distributivity results.
tsc/testdata/baselines/reference/compiler/mappedTypeHomomorphismWithPriorKeyofTChecks1.symbols	Records symbols.
tsc/testdata/baselines/reference/compiler/mappedTypeHomomorphismWithPriorKeyofTChecks1.js	Records emitted output.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[Jake Bailey (jakebailey)]

Can type-fest be changed regardless?

[Mateusz Burzyński (Andarist)]

Can type-fest be changed regardless?

I think so - yes. It's just that I usually prefer to open followup PRs when relevant TS PRs land. It's easier to explain to other maintainers the issue being fixed is not hypothethical. I'll create that type-fest PR in a moment though.