Honor subtyping in generic class match arms' reified type-arg check (B-634) by antoniosarosi · Pull Request #3902 · BoundaryML/baml

antoniosarosi · 2026-07-01T23:32:54Z

Bug

A generic match arm let s: Opt<T> (where T is the enclosing function's type var) lowered its runtime type-test to an exact reified type-arg comparison. When inference pins T to a supertype union of the scrutinee's actual type arg — which happens whenever a default: T argument subtypes, so T reifies to the un-subsumed join Shape | Sq while the value is a plain Opt<Shape> — the exact Shape | Sq == Shape check fails, the arm silently misses, and control falls through to the default.

Repro

interface Shape { function area(self) -> int }
class Sq { s: int, implements Shape { function area(self) -> int { self.s * self.s } } }
class Opt<T> { value: T }
class Non {}

function mk<T>(v: T) -> Opt<T> | Non { Opt<T> { value: v } }

function unwrap_or<T>(o: Opt<T> | Non, default: T) -> T {
  match (o) {
    let s: Opt<T> => s.value,
    _ => default,
  }
}

function bug() -> int {
  let o: Opt<Shape> | Non = mk<Shape>(Sq { s: 5 });
  unwrap_or(o, Sq { s: 0 }).area()   // T infers to `Shape | Sq`
}

bug() returned 0 (expected 25). Widening the default so T pins to Shape directly (let d: Shape = Sq{s:0}) already returned 25. Probing reflect.type_of<T>() inside unwrap_or prints "Shape | Sq", confirming the reified frame type-arg is the un-subsumed join, not Sq.

Root cause

crates/baml_compiler2_mir/src/lower.rs, lower_pattern_test: the typevar-carrying class-pattern branch built the IsType template via ty_to_template, whose TypeVar arm lowers a frame type var to TyTemplate::TypeArgRef(N) (exact). The VM's guard_template_matches (crates/bex_vm/src/vm.rs) checks TypeArgRef(n) with frame_type_args[n] == actual — structural equality. With frame_type_args[0] = Union([Shape, Sq]) and the scrutinee Opt<Shape>'s class_type_args[0] = Shape, Union([Shape, Sq]) == Shape is false, so the arm never matches.

The sibling TypeArgRefOrWildcard(n) arm instead does actual.is_subtype_of(expected) — the correct covariant "does this value belong to type Opt<T>" semantics — and this is exactly what interface class-dispatch guards already use (emit_interface_class_guard_branch, via tir2_to_dispatch_guard_template). Match arms were inconsistently on the exact path.

Fix (MIR layer)

Swap the single ty_to_template call at the typevar class-pattern arm for tir2_to_dispatch_guard_template, so Opt<T> lowers to Class("Opt", [TypeArgRefOrWildcard(0)]). The VM then subtype-checks Shape <: (Shape | Sq) → matches → returns s.value → area() == 25.

This is preferred over broadening the VM's TypeArgRef comparison to is_subtype_of, because it keeps TypeArgRef's exact-materialization meaning intact and makes match/catch class-pattern arms consistent with the already-shipping interface-class-dispatch guard.

Directionality / soundness

The subtype check stays directional: a strictly wider runtime type-arg still fails to match a narrower pinned T ((Shape|Sq).is_subtype_of(Sq) is false). Static typing + covariance already guarantee a well-typed scrutinee's runtime arg is <: the frame T, so this cannot over-match well-typed programs. Concrete parametric arms (Foo<int> vs Foo<string>) carry no enclosing type var and take the untouched exact concrete path.

Tests

bex_vm unit tests on guard_template_matches locking the directional subtype semantics: under-match fix (Shape matches a Shape|Sq frame T), exact TypeArgRef stays exact, a wider runtime arg does not over-match a narrower T, an unknown frame slot is a wildcard, and concrete Foo<int> vs Foo<string> still discriminate.
ns_generic_match_subtype e2e (real VM): narrow-default → 25, widened-default → 25 (control), non-Opt scrutinee falls through to the default, and concrete Foo<int>/Foo<string> still discriminate.
projects/compiles/generic_match_typevar_arm snapshot project locks the emitted template as TypeArgRefOrWildcard (Opt<#0?>, was Opt<#0>) in the MIR/codegen snapshots.

Before / after

function	before	after
`bug2_narrow` (narrow default)	`0`	`25`
`bug2_widened` (widened default, control)	`25`	`25`

Snapshots

No existing e2e/compiled snapshot shifted (verified across the whole baml_src bytecode snapshot: only the new namespace snapshot and the global $init_test registrations appear). New snapshots: the generic_match_typevar_arm compiles project and the generic_match_subtype namespace. The _root.snap change is just the added test registrations.

Local runs green: cargo nextest run -p bex_vm -p baml_compiler2_mir (68 passed) and cargo test -p baml_tests -- --skip parser_stress (2758 passed, 0 failed).

https://linear.app/boundaryml2/issue/B-634

Summary by CodeRabbit

Bug Fixes
- Improved generic match pattern runtime type checks so subtype-based arms continue to match correctly when type inference widens generic arguments.
Tests
- Added end-to-end scenarios for widened generic types, correct arm selection, non-matching fallthrough, and concrete generic discrimination.
- Added unit tests to verify VM dispatch-guard behavior for subtype-or-wildcard vs exact type-template matching.

…B-634) A generic `match` arm `let s: Opt<T>` (where `T` is the enclosing function's type var) lowered its runtime type-test to an EXACT reified type-arg comparison. When inference pins `T` to a supertype union of the scrutinee's actual type arg -- which happens whenever a `default: T` argument subtypes, so `T` reifies to the un-subsumed join `Shape | Sq` while the value is a plain `Opt<Shape>` -- the exact `Shape | Sq == Shape` check failed and the arm silently missed, falling through to the default. Concretely, `unwrap_or(mk<Shape>(Sq{s:5}), Sq{s:0}) .area()` returned 0 instead of 25; widening the default to `Shape` (so `T` pins to `Shape`) already returned 25. Fix (MIR layer): in `lower_pattern_test`, the typevar-carrying class-pattern arm now builds its `IsType` template with `tir2_to_dispatch_guard_template` (frame `TypeVar` -> `TypeArgRefOrWildcard`, subtype-or-wildcard) instead of `ty_to_template` (frame `TypeVar` -> `TypeArgRef`, exact). The VM then compares the reified frame arg with `is_subtype_of` rather than `==`, which is the correct covariant "does this value belong to type `Opt<T>`" semantics and matches the interface class-dispatch guard path (`emit_interface_class_guard_ branch`), which already uses the same template. This keeps `TypeArgRef`'s exact-materialization meaning intact for other uses. Directionality is preserved: a strictly wider runtime type-arg still fails to match a narrower pinned `T`. Concrete parametric arms (`Foo<int>` vs `Foo<string>`) are unaffected -- they carry no enclosing type var and take the exact concrete path. Tests: - bex_vm unit tests on `guard_template_matches` locking the directional subtype semantics (under-match fix, exact `TypeArgRef` stays exact, wider arg does not over-match, unknown-slot wildcard, concrete discrimination). - ns_generic_match_subtype e2e: narrow-default -> 25, widened-default -> 25, non-Opt scrutinee falls through, concrete `Foo<int>`/`Foo<string>` still discriminate. - projects/compiles/generic_match_typevar_arm locks the emitted template as `TypeArgRefOrWildcard` (`Opt<#0?>`) in the MIR/codegen snapshots.

linear · 2026-07-01T23:32:58Z

B-634

vercel · 2026-07-01T23:33:01Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
beps	Ready	Preview, Comment	Jul 1, 2026 11:51pm
promptfiddle	Ready	Preview, Comment	Jul 1, 2026 11:51pm
promptfiddle2	Ready	Preview, Comment	Jul 1, 2026 11:51pm

github-actions · 2026-07-01T23:33:29Z

⏭️ Performance benchmarks were skipped

Perf benchmarks (CodSpeed) are opt-in on pull requests — they no longer run on every push. They always run automatically after merge to canary/main.

To run them on this PR, do any of the following, then push a commit (or re-run CI):

Add RUN_CODSPEED=1 to the PR description, or
Include run-perf or /perf in the PR title or any commit message.

coderabbitai · 2026-07-01T23:34:33Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: b0190197-c79c-499e-a6ec-09a219783e2e

📥 Commits

Reviewing files that changed from the base of the PR and between b82b418 and 57cc6c1.

📒 Files selected for processing (1)

baml_language/crates/bex_vm/src/vm.rs

🚧 Files skipped from review as they are similar to previous changes (1)

baml_language/crates/bex_vm/src/vm.rs

📝 Walkthrough

Walkthrough

Changes the MIR lowering of generic Class(..) pattern tests to use dispatch-guard templates when type variables remain, and adds VM and BAML regression coverage for subtype-or-wildcard matching and concrete type discrimination.

Changes

Generic Match Subtype Dispatch Guard

Layer / File(s)	Summary
Dispatch-guard template for generic type-var pattern tests `baml_language/crates/baml_compiler2_mir/src/lower.rs`	`lower_pattern_test` now uses `tir2_to_dispatch_guard_template` (`TypeArgRefOrWildcard`) instead of `ty_to_template` for `AstPattern::Type` cases where the inferred type is a generic `Class(..)` containing type variables.
VM-level template matching tests `baml_language/crates/bex_vm/src/vm.rs`	New tests verify `TypeArgRefOrWildcard` matches widened frame type-args correctly, while exact `TypeArgRef` and concrete class type-arg templates continue to discriminate strictly.
End-to-end BAML regression fixtures `baml_language/crates/baml_tests/baml_src/ns_generic_match_subtype/generic_match_subtype.baml`, `baml_language/crates/baml_tests/projects/compiles/generic_match_typevar_arm/generic_match_typevar_arm.baml`	New BAML test files define `Shape`/`Sq`/`Opt<T>`/`Foo<T>` types, `match`-based scenarios, and a compile test asserting `IsType` constants render `TypeArgRefOrWildcard` (`#0?`) instead of `TypeArgRef` (`#0`).

Estimated code review effort: 3 (Moderate) | ~25 minutes

Possibly related PRs

BoundaryML/baml#3638: Both PRs adjust generic type-test lowering and dispatch-guard matching so widened type arguments are handled by subtype-or-wildcard semantics.

Poem

A rabbit hopped through typey dust,
Where wildcards matched just as they must.
T grew wide, but still stayed true,
And Sq with Shape both got their due.
Hooray for guards that leap and flow —
Quick ears, keen eyes, and tests that ցույց?

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main fix: subtype-aware reified type-arg checks for generic class match arms.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

Create stacked PR
Commit on current branch

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch antonio/b-634

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

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

coderabbitai

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

baml_language/crates/baml_compiler2_mir/src/lower.rs (1)
15188-15221: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Generic class unions still bypass the template path

baml_language/crates/baml_compiler2_mir/src/lower.rs:15188-15221 only handles top-level Tir2Ty::Class(..). Union-typed patterns like Opt<T> | Other<T> still fall through to the erased runtime lowering, and tir_ty_needs_interface_shape_test only covers Interface members. If those patterns are valid here, they need the same dispatch-guard template path.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@baml_language/crates/baml_compiler2_mir/src/lower.rs` around lines 15188 -
15221, The generic class-pattern lowering only routes top-level Tir2Ty::Class
patterns with typevars through the dispatch-guard template path, so union
patterns like Opt<T> | Other<T> still fall back to erased runtime lowering.
Update the lowering logic in lower.rs around the class-pattern branch to detect
union-typed patterns containing class members with typevars and send them
through tir2_to_dispatch_guard_template and emit_is_type_template_branch as
well, similar to the existing Class(..) handling and the interface guard path.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@baml_language/crates/baml_compiler2_mir/src/lower.rs`:
- Around line 15188-15221: The generic class-pattern lowering only routes
top-level Tir2Ty::Class patterns with typevars through the dispatch-guard
template path, so union patterns like Opt<T> | Other<T> still fall back to
erased runtime lowering. Update the lowering logic in lower.rs around the
class-pattern branch to detect union-typed patterns containing class members
with typevars and send them through tir2_to_dispatch_guard_template and
emit_is_type_template_branch as well, similar to the existing Class(..) handling
and the interface guard path.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 27dde455-9e06-4646-b717-3d1ea6d08df8

📥 Commits

Reviewing files that changed from the base of the PR and between cfafa66 and b82b418.

⛔ Files ignored due to path filters (10)

baml_language/crates/baml_tests/snapshots/baml_src/_root.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/baml_src/generic_match_subtype.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__01_lexer__generic_match_typevar_arm.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__02_parser__generic_match_typevar_arm.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__03_hir.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__04_5_mir.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__04_tir.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__05_diagnostics.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__06_codegen.snap is excluded by !**/*.snap
baml_language/crates/baml_tests/snapshots/compiles/generic_match_typevar_arm/baml_tests__compiles__generic_match_typevar_arm__10_formatter__generic_match_typevar_arm.snap is excluded by !**/*.snap

📒 Files selected for processing (4)

baml_language/crates/baml_compiler2_mir/src/lower.rs
baml_language/crates/baml_tests/baml_src/ns_generic_match_subtype/generic_match_subtype.baml
baml_language/crates/baml_tests/projects/compiles/generic_match_typevar_arm/generic_match_typevar_arm.baml
baml_language/crates/bex_vm/src/vm.rs

github-actions · 2026-07-01T23:42:35Z

Binary size checks passed

✅ 7 passed

	Artifact	Platform	File	Gzip	Gated on	Baseline	Delta	Status
✅	`baml-cli`	Linux	🔒 21.6 MB	9.2 MB	file	21.5 MB	+116.7 KB (+0.5%)	OK
✅	`packed-program`	Linux	🔒 15.8 MB	6.6 MB	file	15.6 MB	+148.6 KB (+1.0%)	OK
✅	`baml-cli`	macOS	🔒 16.6 MB	8.0 MB	file	16.6 MB	+66.3 KB (+0.4%)	OK
✅	`packed-program`	macOS	🔒 12.2 MB	5.8 MB	file	12.1 MB	+83.1 KB (+0.7%)	OK
✅	`baml-cli`	Windows	🔒 18.2 MB	8.2 MB	file	18.1 MB	+107.0 KB (+0.6%)	OK
✅	`packed-program`	Windows	🔒 13.1 MB	5.9 MB	file	13.0 MB	+129.1 KB (+1.0%)	OK
✅	`bridge_wasm`	WASM	14.5 MB	🔒 4.1 MB	gzip	4.1 MB	+44.7 KB (+1.1%)	OK

🔒 = the size this artifact is GATED on (ceiling + delta). Binaries gate on file size (installed binary); WASM gates on gzip (download size). The other size is shown for information only.

Generated by cargo size-gate · workflow run

vercel Bot temporarily deployed to Preview – beps July 1, 2026 23:33 Inactive

coderabbitai Bot reviewed Jul 1, 2026

View reviewed changes

coderabbitai Bot approved these changes Jul 1, 2026

View reviewed changes

vercel Bot temporarily deployed to Preview – promptfiddle2 July 1, 2026 23:39 Inactive

Address clippy lints in B-634 vm.rs type-arg guard tests

57cc6c1

vercel Bot temporarily deployed to Preview – beps July 1, 2026 23:44 Inactive

vercel Bot temporarily deployed to Preview – promptfiddle July 1, 2026 23:44 Inactive

vercel Bot temporarily deployed to Preview – promptfiddle2 July 1, 2026 23:51 Inactive

antoniosarosi added this pull request to the merge queue Jul 2, 2026

Merged via the queue into canary with commit e42faa1 Jul 2, 2026
52 of 53 checks passed

antoniosarosi deleted the antonio/b-634 branch July 2, 2026 00:26

antoniosarosi mentioned this pull request Jul 2, 2026

Fix generic match exhaustiveness so null can't leak into a non-nullable T (B-633) #3904

Closed

coderabbitai Bot mentioned this pull request Jul 7, 2026

Fix matching on lists and maps by element type #3933

Merged

coderabbitai Bot mentioned this pull request Jul 16, 2026

Eliminate parallel type algebra logic in runtime #4046

Merged

Navigation Menu

Repository navigation

Honor subtyping in generic class match arms' reified type-arg check (B-634) - #3902

Honor subtyping in generic class match arms' reified type-arg check (B-634)#3902
antoniosarosi merged 2 commits into
canaryfrom
antonio/b-634

antoniosarosi commented Jul 1, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

linear Bot commented Jul 1, 2026

Uh oh!

vercel Bot commented Jul 1, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jul 1, 2026

Uh oh!

coderabbitai Bot commented Jul 1, 2026 •

edited

Loading

Walkthrough

Changes

Uh oh!

coderabbitai Bot left a comment

Uh oh!

github-actions Bot commented Jul 1, 2026 •

edited

Loading

Uh oh!

Uh oh!

This branch was previously deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

antoniosarosi commented Jul 1, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Bug

Repro

Root cause

Fix (MIR layer)

Directionality / soundness

Tests

Before / after

Snapshots

Summary by CodeRabbit

Uh oh!

linear Bot commented Jul 1, 2026

Uh oh!

vercel Bot commented Jul 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jul 1, 2026

⏭️ Performance benchmarks were skipped

Uh oh!

coderabbitai Bot commented Jul 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jul 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Binary size checks passed

Uh oh!

Uh oh!

This branch was previously deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

antoniosarosi commented Jul 1, 2026 •

edited by coderabbitai Bot

Loading

vercel Bot commented Jul 1, 2026 •

edited

Loading

coderabbitai Bot commented Jul 1, 2026 •

edited

Loading

github-actions Bot commented Jul 1, 2026 •

edited

Loading