Add test for ancestor/descendant type-equality constraints

[cmarcelo]

Add a generics test that exercises equality constraints between nested types and their enclosing type:

• Box == Box.Nested
• Box.Nested == Box

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 0f3652fd-cd99-486d-8d27-1cf197282e2a

📥 Commits

Reviewing files that changed from the base of the PR and between fdab38b and 7ec29c0.

📒 Files selected for processing (1)

• tests/language-feature/generics/type-equality-ancestor-ordering.slang

---

📝 Walkthrough

Walkthrough

Adds a new test that exercises generic type-equality constraints involving an ancestor (nested) relationship, defining struct Box<T> with Nested<U> and two generic functions that assert opposite directions of equality, then invoking both with int and expecting constraint diagnostics.

Changes

Cohort / File(s)	Summary
New Test tests/language-feature/generics/type-equality-ancestor-ordering.slang	Adds test file defining struct Box<T> and nested struct Box<T>::Nested<U>, two generic functions triggerA<X>() where Box<X> == Box<X>.Nested<X> and triggerB<X>() where Box<X>.Nested<X> == Box<X>, plus useIt() invoking both with int and CHECKs for diagnostics.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐇 I hop through boxes, one and two,
Nested loops of type and view,
Equality dreams that twist and bend,
I test them till the errors end,
A rabbit cheers: the checks are through!

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: adding a test for ancestor/descendant type-equality constraints, which is clearly reflected in the file addition and test content.
Description check	✅ Passed	The description is directly related to the changeset, detailing the specific equality constraints being tested (Box == Box.Nested and the opposite direction), which matches the test file content.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (1)

tests/language-feature/generics/type-equality-ancestor-ordering.slang (1)

23-24: Make the two expected failures distinguishable.

These two CHECK_FAIL-DAG lines are identical, so the test can pass without proving that triggerA<int>() and triggerB<int>() each failed independently. Please anchor each expectation to its specific instantiation site, or split the two directions into separate tests.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 14f298e9-d2e6-4e08-af80-1317b32a8eba

📥 Commits

Reviewing files that changed from the base of the PR and between d138e76 and fdab38b.

📒 Files selected for processing (1)

• tests/language-feature/generics/type-equality-ancestor-ordering.slang

[Copilot]

Pull request overview

Adds a new Slang generics regression test to ensure type-equality constraints behave consistently when comparing an enclosing generic type with a nested generic type, regardless of operand order.

Changes:

• Introduces a CHECK_FAIL-based test covering Box<X> == Box<X>.Nested<X> and the reversed constraint.
• Verifies both instantiations emit the expected diagnostic.

[aidanfnv]

Looks good to me

[github-actions]

Review: Add test for ancestor/descendant type-equality constraints

Clean, minimal test-only PR. LGTM ✅

Summary

Adds a single DIAGNOSTIC_TEST:SIMPLE that verifies the compiler correctly rejects unsatisfiable type-equality where constraints between an enclosing generic type (Box<X>) and its nested type (Box<X>.Nested<X>), and that it does so symmetrically regardless of operand order.

Verified

• Diagnostic format: DIAGNOSTIC_TEST:SIMPLE(diag=CHECK): with positional /*CHECK: ... */ blocks — consistent with existing tests (e.g., where-optional-1.slang) and docs/diagnostics.md.
• Error code: E38029 (type-argument-does-not-conform-to-interface) confirmed against slang-diagnostics.lua:3639. Each CHECK block correctly annotates both the error code line and the span message line produced by the diagnostic.
• Caret alignment: 4 spaces + 8 carets (^^^^^^^^) correctly spans the 8-character identifiers triggerA/triggerB at columns 5–12 of the call sites.
• Both orderings tested: triggerA constrains Box<X> == Box<X>.Nested<X>, triggerB constrains the reverse — proving symmetric rejection.
• Positional disambiguation: Each /*CHECK: ... */ block is anchored to its preceding source line, so both calls are independently verified despite producing textually identical diagnostics.
• Exhaustive mode (default): The test cannot pass vacuously — crashes, silent acceptance, or unexpected/missing diagnostics all cause failure.
• No GPU dependency: Diagnostic tests run without hardware, so all CI platforms can validate this.