Repository navigation
Unify TIR Ty into baml_type::Ty - #3722
Merged
Merged
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Contributor
📝 WalkthroughWalkthroughIntroduce a shared type/name crate (qualified names, fresh literals, function-param metadata, runtime-subset conversion, and rendering strategies) and update compiler lowering, MIR/emit, runtime/VM/FFI, tooling, serialization, and tests to use the new APIs and Ty variant shapes. ChangesShared type and rendering
Compiler & TIR
Runtime, VM, FFI, tooling
Tests, snapshots, and wiring
Estimated code review effort: Possibly related PRs:
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
|
⏭️ Performance benchmarks were skippedPerf benchmarks (CodSpeed) are opt-in on pull requests — they no longer run on every push. They always run automatically after merge to To run them on this PR, do any of the following, then push a commit (or re-run CI):
|
2kai2kai2
enabled auto-merge
June 9, 2026 19:51
|
No description provided. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (3)
baml_language/crates/baml_tests/src/compiler2_tir/mod.rs (1)
7-27:⚠️ Potential issue | 🟠 MajorRun and share successful
cargo test --libresults for the Rust change set
Thecargo test --librun did not complete successfully here due to rustup/toolchain download/recovery failures (rename errors) and a subsequentcould not find Cargo.tomlerror (wrong working directory). Re-runcargo test --libfrom the repo root and share the output.🤖 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_tests/src/compiler2_tir/mod.rs` around lines 7 - 27, Re-run the test suite from the repository root by running cargo test --lib and capture the full successful output; the failures reported were due to toolchain/download and wrong-working-directory issues rather than code, so ensure you run from the repo root (not inside crates/baml_tests) and that rustup is healthy, then paste the complete passing test log; refer to the test modules listed in mod.rs (explicit_type_args, inference, phase3a, phase5, phase6, phase7, phase8_exceptions, stream_expansion and the support module) to confirm those tests ran.Source: Coding guidelines
baml_language/crates/bex_external_types/src/host_return.rs (1)
134-143:⚠️ Potential issue | 🟠 Major | ⚡ Quick winHandle
Literal::Bigintin host-return validation.Line 134 currently omits bigint literal equality, so a declared bigint literal return type cannot validate successfully.
💡 Proposed fix
Ty::Literal(lit, _, _) => match (lit, value) { (Literal::Bool(b), BexExternalValue::Bool(v)) => b == v, (Literal::Int(i), BexExternalValue::Int(v)) => i == v, + (Literal::Bigint(i), BexExternalValue::Bigint(v)) => i == v, (Literal::String(s), BexExternalValue::String(v)) => s == v, // `Literal::Float` stores the literal as a string for precision; // match by tag (any float), mirroring // `bex_engine::conversion::value_matches_type`. (Literal::Float(_), BexExternalValue::Float(_)) => true, _ => false, },🤖 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/bex_external_types/src/host_return.rs` around lines 134 - 143, The Ty::Literal match arm in host_return.rs currently ignores bigint literals; add a branch to compare bigint literals by value so declared bigint literal return types validate: in the Ty::Literal(...) => match (lit, value) arm, handle (Literal::Bigint(b), BexExternalValue::Bigint(v)) => b == v (mirror the existing Bool/Int/String patterns), ensuring the Literal::Bigint and BexExternalValue::Bigint symbols are used for the comparison.baml_language/crates/baml_compiler2_mir/src/lower.rs (1)
200-225:⚠️ Potential issue | 🟠 Major | ⚡ Quick winAccount for
Tir2Ty::Functiongenerics in dispatch matching.These helpers still treat two function types as equivalent whenever params/ret/throws line up, but this PR makes function types carry
generic_paramsandgeneric_param_bounds. That means a generic function type can now satisfy, rewrite into, or bind against a monomorphic request (and vice versa), which can select the wrong dispatch candidate or synthesize the wrong guard. Please include the generic head in the match criteria and keep all three helpers aligned.Also applies to: 332-367, 432-457
🤖 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 200 - 225, The function-type comparisons in the three dispatch helpers need to include the function generic head: update the Tir2Ty::Function match arms in tir_type_satisfies_dispatch_request and the two sibling helpers (tir_type_rewrites_dispatch_request and tir_type_binds_dispatch_request) to also check generic_params and generic_param_bounds (e.g., compare lengths and elementwise equality/compatibility of generic params and bounds) as part of the match guard and any subsequent checks; make the generic-head comparison part of the same conjunction that currently checks params.len() and param modes so all three helpers remain aligned when deciding satisfy/rewrites/bind relationships.
🤖 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.
Inline comments:
In `@baml_language/crates/baml_compiler2_tir/src/package_interface.rs`:
- Around line 525-537: build_self_type_for_class currently constructs the
receiver QualifiedTypeName with a hardcoded Name::new("baml"), causing exported
implicit self types to have the wrong package for non-baml packages; update the
construction in build_self_type_for_class to use the actual package identifier
(the package/name value available on the surrounding context or class_data)
instead of the literal "baml" so QualifiedTypeName::new receives the real
package Name, leaving the rest (ns_path, class_data.name, generic args)
unchanged.
- Around line 181-194: The current mismatch comes from ExportedType::Class being
converted to Ty::Class including TypeVar entries while
PackageResolutionContext::resolve_type() (via def_to_ty()) emits own-package
generic classes with an empty vec, causing asymmetric Ty shapes; update
def_to_ty() (or have resolve_type call the same conversion path) to include the
declaring type's generic parameters as Ty::TypeVar entries (using the same
TyAttr::default() pattern) so that both dependency and own-package generic
classes produce Ty::Class(qtn, vec![Ty::TypeVar(...) ...], TyAttr::default())
consistently; ensure you reference ExportedType::Class, Ty::Class,
PackageResolutionContext::resolve_type(), and def_to_ty() when making the
change.
In `@baml_language/crates/bex_vm/src/package_baml/json.rs`:
- Around line 559-574: The comment and error classification are wrong:
Ty::RustType and Ty::Type are runtime leaf types per Ty::validate_runtime() and
should not be described as “compiler-only”; update the serialize and deserialize
match arms in the JSON handling to either move Ty::RustType and Ty::Type out of
the “TIR-internal compiler-only types” group or split them into their own arm,
and change the raise_serialize/raise_deserialize messages to indicate “not
JSON-representable” (or similar) instead of “compiler_only”; refer to the Ty
enum variants Ty::RustType and Ty::Type, the validate_runtime() caller for
intended semantics, and the raise_serialize call sites in the
serialize/deserialize blocks to make the error text and grouping accurate.
---
Outside diff comments:
In `@baml_language/crates/baml_compiler2_mir/src/lower.rs`:
- Around line 200-225: The function-type comparisons in the three dispatch
helpers need to include the function generic head: update the Tir2Ty::Function
match arms in tir_type_satisfies_dispatch_request and the two sibling helpers
(tir_type_rewrites_dispatch_request and tir_type_binds_dispatch_request) to also
check generic_params and generic_param_bounds (e.g., compare lengths and
elementwise equality/compatibility of generic params and bounds) as part of the
match guard and any subsequent checks; make the generic-head comparison part of
the same conjunction that currently checks params.len() and param modes so all
three helpers remain aligned when deciding satisfy/rewrites/bind relationships.
In `@baml_language/crates/baml_tests/src/compiler2_tir/mod.rs`:
- Around line 7-27: Re-run the test suite from the repository root by running
cargo test --lib and capture the full successful output; the failures reported
were due to toolchain/download and wrong-working-directory issues rather than
code, so ensure you run from the repo root (not inside crates/baml_tests) and
that rustup is healthy, then paste the complete passing test log; refer to the
test modules listed in mod.rs (explicit_type_args, inference, phase3a, phase5,
phase6, phase7, phase8_exceptions, stream_expansion and the support module) to
confirm those tests ran.
In `@baml_language/crates/bex_external_types/src/host_return.rs`:
- Around line 134-143: The Ty::Literal match arm in host_return.rs currently
ignores bigint literals; add a branch to compare bigint literals by value so
declared bigint literal return types validate: in the Ty::Literal(...) => match
(lit, value) arm, handle (Literal::Bigint(b), BexExternalValue::Bigint(v)) => b
== v (mirror the existing Bool/Int/String patterns), ensuring the
Literal::Bigint and BexExternalValue::Bigint symbols are used for the
comparison.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: 3b8c1877-0318-4b4d-b3d0-be947ae7cdba
⛔ Files ignored due to path filters (1)
baml_language/Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (54)
baml_language/Cargo.tomlbaml_language/crates/baml_base/src/core_types.rsbaml_language/crates/baml_compiler2_emit/src/emit.rsbaml_language/crates/baml_compiler2_emit/src/lib.rsbaml_language/crates/baml_compiler2_emit/src/stack_carry.rsbaml_language/crates/baml_compiler2_mir/src/lib.rsbaml_language/crates/baml_compiler2_mir/src/lower.rsbaml_language/crates/baml_compiler2_tir/src/associated_projection.rsbaml_language/crates/baml_compiler2_tir/src/builder.rsbaml_language/crates/baml_compiler2_tir/src/exhaustiveness.rsbaml_language/crates/baml_compiler2_tir/src/generics.rsbaml_language/crates/baml_compiler2_tir/src/infer_context.rsbaml_language/crates/baml_compiler2_tir/src/interfaces.rsbaml_language/crates/baml_compiler2_tir/src/lower_type_expr.rsbaml_language/crates/baml_compiler2_tir/src/narrowing.rsbaml_language/crates/baml_compiler2_tir/src/normalize.rsbaml_language/crates/baml_compiler2_tir/src/package_interface.rsbaml_language/crates/baml_compiler2_tir/src/throw_inference.rsbaml_language/crates/baml_compiler2_tir/src/throws_analysis.rsbaml_language/crates/baml_compiler2_tir/src/ty.rsbaml_language/crates/baml_exec/src/auto_cli.rsbaml_language/crates/baml_exec/src/clap_target.rsbaml_language/crates/baml_lsp2_actions/src/completions.rsbaml_language/crates/baml_lsp2_actions/src/describe.rsbaml_language/crates/baml_lsp2_actions/src/type_info.rsbaml_language/crates/baml_lsp2_actions/src/utils.rsbaml_language/crates/baml_project/src/client_codegen.rsbaml_language/crates/baml_tests/src/compiler2_tir/inference.rsbaml_language/crates/baml_tests/src/compiler2_tir/mod.rsbaml_language/crates/baml_tests/src/compiler2_tir/phase6.rsbaml_language/crates/baml_tests/tests/exceptions.rsbaml_language/crates/baml_type/Cargo.tomlbaml_language/crates/baml_type/src/lib.rsbaml_language/crates/baml_type/src/names.rsbaml_language/crates/baml_type/src/primitive.rsbaml_language/crates/baml_type/src/runtime_ty.rsbaml_language/crates/baml_type/src/simplify_sap.rsbaml_language/crates/baml_type/src/template.rsbaml_language/crates/baml_type/tests/simplify_sap_tests.rsbaml_language/crates/bex_engine/src/conversion.rsbaml_language/crates/bex_engine/src/lib.rsbaml_language/crates/bex_external_types/src/host_return.rsbaml_language/crates/bex_heap/src/accessor.rsbaml_language/crates/bex_heap/src/gc.rsbaml_language/crates/bex_sap/src/sap_model/convert.rsbaml_language/crates/bex_sap/src/to_external.rsbaml_language/crates/bex_vm/src/package_baml/json.rsbaml_language/crates/bex_vm/src/package_baml/mod.rsbaml_language/crates/bex_vm/src/package_baml/type_class.rsbaml_language/crates/bex_vm/src/package_baml/unstable.rsbaml_language/crates/bridge_ctypes/src/value_encode.rsbaml_language/crates/sys_llm/src/lib.rsbaml_language/crates/sys_llm/src/types/output_format.rsbaml_language/crates/tools_sap_visualizer/src/compile.rs
This is in preparation for larger changes. Also adds `subenum` for `baml_type::Ty` (will be extended by further changes)
Unifies a lot of behavior.
It is not a part of the name and was only used in a few places. We should just use the type itself to get generic params, not the QTE.
For eventually using it in TIR
- Clean up `qtn_to_type_name` now that they are unified - Replace `Ty::Opaque` with real variants
We actually need them to be separate so that `RuntimeTy` can be deeply `RuntimeTy`. We still use `subenum` on `RuntimeTy`.
2kai2kai2
force-pushed
the
kai/tir-cleanup
branch
from
June 9, 2026 20:22
854698e to
966a336
Compare
2kai2kai2
force-pushed
the
kai/tir-cleanup
branch
from
June 9, 2026 20:26
966a336 to
89b1f6d
Compare
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (3)
baml_language/crates/baml_compiler2_mir/src/lower.rs (1)
1836-1843:⚠️ Potential issue | 🟠 Major | ⚡ Quick winDon't hardcode
baml.iterin the generic registry lookup path.Line 1841 now forces package discovery to use
baml.iter.<name>, butinterface_view_from_registryis still used for arbitrarytarget_tns viainterface_view_for_tir_ty. That can make blanket-impl lookup miss the real interface package for non-baml.iterinterfaces and fail to resolve the interface view.Suggested fix
- let target_qtn = Self::baml_iter_qtn(target_tn.name().as_str()); + let target_qtn = target_tn.clone();🤖 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 1836 - 1843, The code currently forces discovery to use baml.iter by calling Self::baml_iter_qtn(...) inside interface_view_from_registry; change it to use the actual target type name instead of hardcoding baml.iter. Specifically, stop calling baml_iter_qtn and pass the real qualified type name for target_tn into registry_package_ids_for_interface_lookup (or derive the qtn from target_tn directly) so package lookup for interface_view_from_registry and interface_view_for_tir_ty can find non-baml.iter interface packages; update references around interface_view_from_registry, target_tn, and registry_package_ids_for_interface_lookup accordingly.baml_language/crates/baml_compiler2_tir/src/package_interface.rs (2)
237-249:⚠️ Potential issue | 🟠 Major | ⚡ Quick winPreserve class generics when rewriting
Self.
substitute_self_in()is fed a barePath(Foo)here, so an explicitSelfinsideclass Box<T>lowers toTy::Class(..., []), while the implicitselfparameter now comes frombuild_self_type_for_class()asTy::Class(..., [TypeVar(T)]). That gives the same receiver type two differentTyshapes inside one method signature and will break equality/rendering for generic methods. Please carryclass_data.generic_paramsthrough theSelfreplacement path too.Also applies to: 257-261
🤖 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_tir/src/package_interface.rs` around lines 237 - 249, The Self substitution currently uses a bare path via type_expr_for_name(class_data.name) which loses class generic parameters; change the self_replacement passed to substitute_self_in to include the class's generic params (mirror what build_self_type_for_class produces) by constructing a TypeExpr/Path for the enclosing class that carries class_data.generic_params as type arguments (or corresponding TypeVar expressions) before calling substitute_self_in in the lower_with_self closure, and make the same change at the other substitute_self_in call around lines 257-261 so explicit Self and the implicit receiver share the same Ty shape for generics.
922-927:⚠️ Potential issue | 🟠 Major | ⚡ Quick winOwn-package generic interfaces still erase their type variables.
Definition::Interface(_)still constructsTy::Interface(..., vec![], ...), even though this PR moved unspecialized generic classes toTypeVarargs. A same-package generic interface resolved throughPackageResolutionContext::resolve_type()will therefore still come back asIfaceinstead ofIface<T>, keeping interface resolution inconsistent with the newTycontract.🤖 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_tir/src/package_interface.rs` around lines 922 - 927, The Interface arm in the Definition::Interface match currently constructs Ty::Interface with an empty type-arg vector, which erases same-package generic type parameters; update that construction in package_interface.rs (the Definition::Interface(_) match) to populate the type argument vector from the interface's generic/type parameters (mapping each parameter to an unspecialized Ty::TypeVar or whatever the project uses for unspecialized generics) so that Ty::Interface(qualify_def(db, def, &name), vec_of_typevars, ...) preserves the interface's type variables and stays consistent with PackageResolutionContext::resolve_type and the new Ty contract.
🤖 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.
Inline comments:
In `@baml_language/crates/baml_type/src/lib.rs`:
- Around line 154-157: validate_runtime() currently allows Ty::Literal(_,
Freshness::Fresh, _) through, which leaks the compiler-only Freshness into
runtime equality/hashing/serialization; change validate_runtime() to normalize
any Ty::Literal to use Freshness::Regular (erase Freshness) before
accepting/returning it so runtime representations always use Regular. Locate the
handling of the Ty::Literal variant in validate_runtime() and replace/augment
that branch to construct/return a Literal with Freshness::Regular (or otherwise
map Freshness::Fresh -> Freshness::Regular) so runtime checks and Borsh
serialization cannot observe Freshness.
- Around line 903-910: render_user_facing currently bypasses two user-facing
policies when calling render_with with CanonicalTyRender: it prints raw generic
parameter names instead of using the serializer's user-facing alias and it
renders the no-throws sentinel as "void" instead of the user-facing "never".
Update the CanonicalTyRender path used by render_user_facing (and the code paths
around the generic param and effect rendering) to call the same user-facing
helpers as canonical rendering: use s.type_var(...) (or the equivalent
user-facing serializer method) to render generic/type variables and map the
no-throws sentinel to "never" in the effect-rendering logic so user-facing
strings never expose __effect_param_N or produce "throws void"; keep the caller
as render_user_facing -> render_with(&CanonicalTyRender { user_facing: true })
but ensure the internals honor user_facing when choosing s.type_var and the
no-throws representation.
- Around line 173-179: The runtime validator is currently skipping
Function.generic_param_bounds so compiler-only sentinels (Unknown, Error,
Evolving*, etc.) can leak into runtime metadata; update the runtime validation
to walk and validate each entry in generic_param_bounds (in addition to
generic_params and params) — either by calling a new metadata-aware validator
(e.g., validate_runtime_metadata_ty) or by extending validate_runtime() with a
flag to allow/handle reflective-only forms like TypeVar and
AssociatedTypeProjection, and ensure Function, generic_param_bounds, TypeVar,
and AssociatedTypeProjection are explicitly handled so sentinel types are
rejected or normalized before emitting reflective payloads.
In `@baml_language/crates/baml_type/src/primitive.rs`:
- Around line 7-20: Ty::is_primitive currently omits the new media variants so
Image/Audio/Video/Pdf are misclassified; update the is_primitive matcher in the
Ty impl (function Ty::is_primitive) to include the media branch (e.g., add |
Ty::Media(..) | to the matches list) and likewise update any other similar
primitive-check helpers (the other is_primitive-like match blocks referenced in
the comment) so media types are treated as primitives.
---
Outside diff comments:
In `@baml_language/crates/baml_compiler2_mir/src/lower.rs`:
- Around line 1836-1843: The code currently forces discovery to use baml.iter by
calling Self::baml_iter_qtn(...) inside interface_view_from_registry; change it
to use the actual target type name instead of hardcoding baml.iter.
Specifically, stop calling baml_iter_qtn and pass the real qualified type name
for target_tn into registry_package_ids_for_interface_lookup (or derive the qtn
from target_tn directly) so package lookup for interface_view_from_registry and
interface_view_for_tir_ty can find non-baml.iter interface packages; update
references around interface_view_from_registry, target_tn, and
registry_package_ids_for_interface_lookup accordingly.
In `@baml_language/crates/baml_compiler2_tir/src/package_interface.rs`:
- Around line 237-249: The Self substitution currently uses a bare path via
type_expr_for_name(class_data.name) which loses class generic parameters; change
the self_replacement passed to substitute_self_in to include the class's generic
params (mirror what build_self_type_for_class produces) by constructing a
TypeExpr/Path for the enclosing class that carries class_data.generic_params as
type arguments (or corresponding TypeVar expressions) before calling
substitute_self_in in the lower_with_self closure, and make the same change at
the other substitute_self_in call around lines 257-261 so explicit Self and the
implicit receiver share the same Ty shape for generics.
- Around line 922-927: The Interface arm in the Definition::Interface match
currently constructs Ty::Interface with an empty type-arg vector, which erases
same-package generic type parameters; update that construction in
package_interface.rs (the Definition::Interface(_) match) to populate the type
argument vector from the interface's generic/type parameters (mapping each
parameter to an unspecialized Ty::TypeVar or whatever the project uses for
unspecialized generics) so that Ty::Interface(qualify_def(db, def, &name),
vec_of_typevars, ...) preserves the interface's type variables and stays
consistent with PackageResolutionContext::resolve_type and the new Ty contract.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: 24a0cdf1-f5dd-48b1-ad27-d21730062c74
⛔ Files ignored due to path filters (5)
baml_language/Cargo.lockis excluded by!**/*.lockbaml_language/crates/baml_tests/snapshots/diagnostic_errors/generic_class_destructure_no_type_args/baml_tests__diagnostic_errors__generic_class_destructure_no_type_args__04_tir.snapis excluded by!**/*.snapbaml_language/crates/baml_tests/snapshots/diagnostic_errors/generic_class_destructure_no_type_args/baml_tests__diagnostic_errors__generic_class_destructure_no_type_args__05_diagnostics.snapis excluded by!**/*.snapbaml_language/crates/baml_tests/snapshots/diagnostic_errors/patterns_class_destructure/baml_tests__diagnostic_errors__patterns_class_destructure__04_tir.snapis excluded by!**/*.snapbaml_language/crates/baml_tests/snapshots/diagnostic_errors/patterns_class_destructure/baml_tests__diagnostic_errors__patterns_class_destructure__05_diagnostics.snapis excluded by!**/*.snap
📒 Files selected for processing (54)
baml_language/Cargo.tomlbaml_language/crates/baml_base/src/core_types.rsbaml_language/crates/baml_compiler2_emit/src/emit.rsbaml_language/crates/baml_compiler2_emit/src/lib.rsbaml_language/crates/baml_compiler2_emit/src/stack_carry.rsbaml_language/crates/baml_compiler2_mir/src/lib.rsbaml_language/crates/baml_compiler2_mir/src/lower.rsbaml_language/crates/baml_compiler2_tir/src/associated_projection.rsbaml_language/crates/baml_compiler2_tir/src/builder.rsbaml_language/crates/baml_compiler2_tir/src/exhaustiveness.rsbaml_language/crates/baml_compiler2_tir/src/generics.rsbaml_language/crates/baml_compiler2_tir/src/infer_context.rsbaml_language/crates/baml_compiler2_tir/src/interfaces.rsbaml_language/crates/baml_compiler2_tir/src/lower_type_expr.rsbaml_language/crates/baml_compiler2_tir/src/narrowing.rsbaml_language/crates/baml_compiler2_tir/src/normalize.rsbaml_language/crates/baml_compiler2_tir/src/package_interface.rsbaml_language/crates/baml_compiler2_tir/src/throw_inference.rsbaml_language/crates/baml_compiler2_tir/src/throws_analysis.rsbaml_language/crates/baml_compiler2_tir/src/ty.rsbaml_language/crates/baml_exec/src/auto_cli.rsbaml_language/crates/baml_exec/src/clap_target.rsbaml_language/crates/baml_lsp2_actions/src/completions.rsbaml_language/crates/baml_lsp2_actions/src/describe.rsbaml_language/crates/baml_lsp2_actions/src/type_info.rsbaml_language/crates/baml_lsp2_actions/src/utils.rsbaml_language/crates/baml_project/src/client_codegen.rsbaml_language/crates/baml_tests/src/compiler2_tir/inference.rsbaml_language/crates/baml_tests/src/compiler2_tir/mod.rsbaml_language/crates/baml_tests/src/compiler2_tir/phase6.rsbaml_language/crates/baml_tests/tests/exceptions.rsbaml_language/crates/baml_type/Cargo.tomlbaml_language/crates/baml_type/src/lib.rsbaml_language/crates/baml_type/src/names.rsbaml_language/crates/baml_type/src/primitive.rsbaml_language/crates/baml_type/src/runtime_ty.rsbaml_language/crates/baml_type/src/simplify_sap.rsbaml_language/crates/baml_type/src/template.rsbaml_language/crates/baml_type/tests/simplify_sap_tests.rsbaml_language/crates/bex_engine/src/conversion.rsbaml_language/crates/bex_engine/src/lib.rsbaml_language/crates/bex_external_types/src/host_return.rsbaml_language/crates/bex_heap/src/accessor.rsbaml_language/crates/bex_heap/src/gc.rsbaml_language/crates/bex_sap/src/sap_model/convert.rsbaml_language/crates/bex_sap/src/to_external.rsbaml_language/crates/bex_vm/src/package_baml/json.rsbaml_language/crates/bex_vm/src/package_baml/mod.rsbaml_language/crates/bex_vm/src/package_baml/type_class.rsbaml_language/crates/bex_vm/src/package_baml/unstable.rsbaml_language/crates/bridge_ctypes/src/value_encode.rsbaml_language/crates/sys_llm/src/lib.rsbaml_language/crates/sys_llm/src/types/output_format.rsbaml_language/crates/tools_sap_visualizer/src/compile.rs
✅ Files skipped from review due to trivial changes (7)
- baml_language/crates/baml_compiler2_tir/src/infer_context.rs
- baml_language/crates/baml_exec/src/clap_target.rs
- baml_language/crates/baml_compiler2_mir/src/lib.rs
- baml_language/crates/baml_lsp2_actions/src/describe.rs
- baml_language/crates/baml_tests/src/compiler2_tir/phase6.rs
- baml_language/crates/baml_compiler2_tir/src/associated_projection.rs
- baml_language/crates/bex_heap/src/gc.rs
🚧 Files skipped from review as they are similar to previous changes (39)
- baml_language/Cargo.toml
- baml_language/crates/baml_type/Cargo.toml
- baml_language/crates/bex_vm/src/package_baml/mod.rs
- baml_language/crates/baml_tests/tests/exceptions.rs
- baml_language/crates/baml_base/src/core_types.rs
- baml_language/crates/bex_sap/src/to_external.rs
- baml_language/crates/baml_compiler2_tir/src/throws_analysis.rs
- baml_language/crates/bex_engine/src/lib.rs
- baml_language/crates/baml_type/tests/simplify_sap_tests.rs
- baml_language/crates/baml_project/src/client_codegen.rs
- baml_language/crates/baml_lsp2_actions/src/type_info.rs
- baml_language/crates/tools_sap_visualizer/src/compile.rs
- baml_language/crates/baml_type/src/simplify_sap.rs
- baml_language/crates/baml_type/src/template.rs
- baml_language/crates/baml_compiler2_emit/src/stack_carry.rs
- baml_language/crates/baml_lsp2_actions/src/utils.rs
- baml_language/crates/bex_heap/src/accessor.rs
- baml_language/crates/bex_vm/src/package_baml/type_class.rs
- baml_language/crates/baml_compiler2_tir/src/throw_inference.rs
- baml_language/crates/baml_compiler2_emit/src/emit.rs
- baml_language/crates/baml_compiler2_tir/src/ty.rs
- baml_language/crates/baml_tests/src/compiler2_tir/mod.rs
- baml_language/crates/baml_lsp2_actions/src/completions.rs
- baml_language/crates/bridge_ctypes/src/value_encode.rs
- baml_language/crates/baml_compiler2_tir/src/generics.rs
- baml_language/crates/baml_type/src/runtime_ty.rs
- baml_language/crates/bex_engine/src/conversion.rs
- baml_language/crates/baml_compiler2_emit/src/lib.rs
- baml_language/crates/sys_llm/src/lib.rs
- baml_language/crates/bex_external_types/src/host_return.rs
- baml_language/crates/baml_compiler2_tir/src/interfaces.rs
- baml_language/crates/baml_type/src/names.rs
- baml_language/crates/bex_sap/src/sap_model/convert.rs
- baml_language/crates/baml_compiler2_tir/src/exhaustiveness.rs
- baml_language/crates/bex_vm/src/package_baml/json.rs
- baml_language/crates/sys_llm/src/types/output_format.rs
- baml_language/crates/baml_compiler2_tir/src/narrowing.rs
- baml_language/crates/baml_compiler2_tir/src/normalize.rs
- baml_language/crates/baml_compiler2_tir/src/builder.rs
This was referenced Jul 13, 2026
This branch was previously deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Type representations for both the compiler and VM are now unified into a single location. We also use the subenum crate to represent subsets of the type system, so we can make invalid states unrepresentable in various contexts.
Rollout of the runtime/subenum types will come in a followup PR.
Summary by CodeRabbit