bxff:fix-ty-2427 - Branch - astral-sh/ruff

[ty] Fix contravariant type variable bound checking in specialization inference

#22488

Comparing

bxff:fix-ty-2427

(

b9adfa7

) with

main

(

11cc324

)

-1%

Untouched: 23

Skipped: 30

Benchmarks

Skipped (30)

Passed

multithreaded

crates/ruff_benchmark/benches/ty_walltime.rs

+2%

1.1 s1.1 s

ty_check_file[incremental]

crates/ruff_benchmark/benches/ty.rs::check_file::benchmark_incremental

6.1 ms6.1 ms

tanjun

crates/ruff_benchmark/benches/ty_walltime.rs

2.5 s2.5 s

sympy

crates/ruff_benchmark/benches/ty_walltime.rs

50.5 s50.6 s

static_frame

crates/ruff_benchmark/benches/ty_walltime.rs

21 s21 s

altair

crates/ruff_benchmark/benches/ty_walltime.rs

4.5 s4.6 s

freqtrade

crates/ruff_benchmark/benches/ty_walltime.rs

7.8 s7.8 s

ty_micro[many_enum_members_2]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_many_enum_members_2

-1%

152.2 ms153.1 ms

pandas

crates/ruff_benchmark/benches/ty_walltime.rs

-1%

62.3 s62.7 s

ty_micro[many_enum_members]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_many_enum_members

-1%

122.9 ms123.6 ms

hydra-zen

crates/ruff_benchmark/benches/ty.rs::project::hydra::project

-1%

1.2 s1.2 s

ty_check_file[cold]

crates/ruff_benchmark/benches/ty.rs::check_file::benchmark_cold

-1%

130.2 ms131.3 ms

ty_micro[many_string_assignments]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_many_string_assignments

-1%

77.9 ms78.7 ms

anyio

crates/ruff_benchmark/benches/ty.rs::project::anyio::project

-1%

1.2 s1.2 s

ty_micro[many_tuple_assignments]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_many_tuple_assignments

-1%

65.5 ms66.3 ms

attrs

crates/ruff_benchmark/benches/ty.rs::project::attrs::project

-1%

428 ms433.3 ms

ty_micro[complex_constrained_attributes_3]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_complex_constrained_attributes_3

-1%

70.8 ms71.7 ms

ty_micro[many_tuple_assignments]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_tuple_implicit_instance_attributes

-1%

67.6 ms68.5 ms

ty_micro[complex_constrained_attributes_2]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_complex_constrained_attributes_2

-1%

67.6 ms68.4 ms

ty_micro[complex_constrained_attributes_1]

crates/ruff_benchmark/benches/ty.rs::micro::benchmark_complex_constrained_attributes_1

-1%

67.6 ms68.4 ms

pydantic

crates/ruff_benchmark/benches/ty_walltime.rs

-1%

10 s10.2 s

colour_science

crates/ruff_benchmark/benches/ty_walltime.rs

-2%

86.6 s88.1 s

DateType

crates/ruff_benchmark/benches/ty.rs::project::datetype::project

-2%

231.4 ms235.8 ms

Commits

Click on a commit to change the comparison range

Base

main

11cc324

-0.74%

[ty] Fix contravariant type variable bound checking in specialization inference ## Summary Correctly handle upper bounds for contravariant type variables during specialization inference. Previously, the type checker incorrectly applied covariant subtyping rules, requiring the actual type to directly satisfy the bound rather than checking for a valid intersection. In contravariant positions, subtyping relationships are inverted. The bug caused valid code like `f(x: Contra[str])` where `f` expects `Contra[T: int]` to be incorrectly rejected, when it should solve `T` to `Never` (the intersection of `int` and `str`). Closes https://github.com/astral-sh/ty/issues/2427 ## Details - Added `is_contravariant()` helper to `TypeVarVariance` in `variance.rs` - Updated `SpecializationBuilder::infer_map_impl` in `generics.rs` to treat bounds and constraints differently based on variance: * Skip immediate `ty <: bound` check for contravariant upper bounds * Flip constraint check to `constraint <: ty` for contravariant positions - Added test case for bounded contravariant type variables in `variance.md` - All 308 mdtest cases pass & 150 ty_python_semantic unit tests pass

b9adfa7

18 hours ago

by bxff

Home Terms Privacy Docs