[ty] support generic aliases in type[...], like type[C[int]]
#21552
+82
−10
Conversation
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
oconnor663
requested review from
AlexWaygood,
carljm,
dcreager and
sharkdp
as code owners
Contributor
Author
|
Also thanks @carljm for talking me through the implementation of this :) |
Diagnostic diff on typing conformance testsChanges were detected when running ty on typing conformance tests--- old-output.txt 2025-11-21 21:46:55.218257234 +0000
+++ new-output.txt 2025-11-21 21:46:58.715273992 +0000
@@ -220,6 +220,7 @@
constructors_call_new.py:117:1: error[type-assertion-failure] Type `Class8[list[str]]` does not match asserted type `Class8[str]`
constructors_call_new.py:125:42: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `Self@__new__`
constructors_call_new.py:140:47: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `Class11[int]`
+constructors_call_new.py:145:1: error[invalid-argument-type] Argument to function `__new__` is incorrect: Expected `type[Class11[int]]`, found `<class 'Class11[str]'>`
constructors_call_type.py:40:5: error[missing-argument] No arguments provided for required parameters `x`, `y` of function `__new__`
constructors_call_type.py:50:5: error[missing-argument] No arguments provided for required parameters `x`, `y` of bound method `__init__`
constructors_call_type.py:59:9: error[too-many-positional-arguments] Too many positional arguments to bound method `__init__`: expected 1, got 2
@@ -415,25 +416,26 @@
generics_basic.py:171:1: error[invalid-generic-class] `Generic` base class must include all type variables used in other base classes
generics_basic.py:172:1: error[invalid-generic-class] `Generic` base class must include all type variables used in other base classes
generics_basic.py:199:5: error[type-assertion-failure] Type `Iterator[Any]` does not match asserted type `Unknown`
-generics_defaults.py:30:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'NoNonDefaults'>`
-generics_defaults.py:31:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'NoNonDefaults[str, int]'>`
-generics_defaults.py:32:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'NoNonDefaults[str, int]'>`
-generics_defaults.py:38:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'OneDefault[int | float, bool]'>`
-generics_defaults.py:45:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults'>`
-generics_defaults.py:46:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
+generics_defaults.py:30:1: error[type-assertion-failure] Type `type[NoNonDefaults[str, int]]` does not match asserted type `<class 'NoNonDefaults'>`
+generics_defaults.py:31:1: error[type-assertion-failure] Type `type[NoNonDefaults[str, int]]` does not match asserted type `<class 'NoNonDefaults[str, int]'>`
+generics_defaults.py:32:1: error[type-assertion-failure] Type `type[NoNonDefaults[str, int]]` does not match asserted type `<class 'NoNonDefaults[str, int]'>`
+generics_defaults.py:38:1: error[type-assertion-failure] Type `type[OneDefault[int | float, bool]]` does not match asserted type `<class 'OneDefault[int | float, bool]'>`
+generics_defaults.py:45:1: error[type-assertion-failure] Type `type[AllTheDefaults[Any, Any, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults'>`
+generics_defaults.py:46:1: error[type-assertion-failure] Type `type[AllTheDefaults[int, int | float | complex, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
generics_defaults.py:50:1: error[missing-argument] No argument provided for required parameter `T2` of class `AllTheDefaults`
-generics_defaults.py:52:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
-generics_defaults.py:55:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
-generics_defaults.py:59:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
-generics_defaults.py:63:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
-generics_defaults.py:79:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'Class_ParamSpec'>`
+generics_defaults.py:52:1: error[type-assertion-failure] Type `type[AllTheDefaults[int, int | float | complex, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
+generics_defaults.py:55:1: error[type-assertion-failure] Type `type[AllTheDefaults[int, int | float | complex, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
+generics_defaults.py:59:1: error[type-assertion-failure] Type `type[AllTheDefaults[int, int | float | complex, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
+generics_defaults.py:63:1: error[type-assertion-failure] Type `type[AllTheDefaults[int, int | float | complex, str, int, bool]]` does not match asserted type `<class 'AllTheDefaults[int, int | float | complex, str, int, bool]'>`
+generics_defaults.py:79:1: error[type-assertion-failure] Type `Unknown` does not match asserted type `<class 'Class_ParamSpec'>`
+generics_defaults.py:79:35: error[too-many-positional-arguments] Too many positional arguments to class `Class_ParamSpec`: expected 1, got 2
generics_defaults.py:80:1: error[type-assertion-failure] Type `Unknown` does not match asserted type `Class_ParamSpec[Unknown]`
generics_defaults.py:80:32: error[too-many-positional-arguments] Too many positional arguments to class `Class_ParamSpec`: expected 1, got 2
generics_defaults.py:81:1: error[type-assertion-failure] Type `Unknown` does not match asserted type `Class_ParamSpec[Unknown]`
generics_defaults.py:81:29: error[invalid-type-form] List literals are not allowed in this context in a type expression: Did you mean `tuple[bool, bool]`?
generics_defaults.py:81:46: error[too-many-positional-arguments] Too many positional arguments to class `Class_ParamSpec`: expected 1, got 2
generics_defaults.py:91:26: error[invalid-argument-type] `@Todo(starred expression)` is not a valid argument to `Generic`
-generics_defaults.py:94:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'Class_TypeVarTuple'>`
+generics_defaults.py:94:1: error[type-assertion-failure] Type `@Todo(specialized non-generic class)` does not match asserted type `<class 'Class_TypeVarTuple'>`
generics_defaults.py:95:1: error[type-assertion-failure] Type `@Todo(specialized non-generic class)` does not match asserted type `Class_TypeVarTuple`
generics_defaults.py:127:32: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `T4@func1`
generics_defaults.py:131:1: error[type-assertion-failure] Type `Any` does not match asserted type `int`
@@ -442,15 +444,15 @@
generics_defaults.py:155:49: error[invalid-type-form] List literals are not allowed in this context in a type expression: Did you mean `tuple[int | float, bool]`?
generics_defaults.py:156:58: error[invalid-type-form] List literals are not allowed in this context in a type expression: Did you mean `list[bytes]`?
generics_defaults.py:170:1: error[type-assertion-failure] Type `(Foo7[int], /) -> Foo7[int]` does not match asserted type `def meth(self, /) -> Self@meth`
-generics_defaults_referential.py:23:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'slice'>`
+generics_defaults_referential.py:23:1: error[type-assertion-failure] Type `type[slice[int, int, int | None]]` does not match asserted type `<class 'slice'>`
generics_defaults_referential.py:36:13: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int`, found `Literal[""]`
generics_defaults_referential.py:37:10: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int`, found `Literal[""]`
-generics_defaults_referential.py:94:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'Bar'>`
-generics_defaults_referential.py:95:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'Bar[int, list[int]]'>`
+generics_defaults_referential.py:94:1: error[type-assertion-failure] Type `type[Bar[Any, list[Any]]]` does not match asserted type `<class 'Bar'>`
+generics_defaults_referential.py:95:1: error[type-assertion-failure] Type `type[Bar[int, list[int]]]` does not match asserted type `<class 'Bar[int, list[int]]'>`
generics_defaults_specialization.py:26:5: error[type-assertion-failure] Type `SomethingWithNoDefaults[int, str]` does not match asserted type `SomethingWithNoDefaults[int, typing.TypeVar]`
generics_defaults_specialization.py:27:5: error[type-assertion-failure] Type `SomethingWithNoDefaults[int, bool]` does not match asserted type `@Todo(specialized generic alias in type expression)`
generics_defaults_specialization.py:30:1: error[non-subscriptable] Cannot subscript object of type `<class 'SomethingWithNoDefaults[int, typing.TypeVar]'>` with no `__class_getitem__` method
-generics_defaults_specialization.py:45:1: error[type-assertion-failure] Type `@Todo(unsupported nested subscript in type[X])` does not match asserted type `<class 'Bar'>`
+generics_defaults_specialization.py:45:1: error[type-assertion-failure] Type `type[Bar[str]]` does not match asserted type `<class 'Bar'>`
generics_paramspec_basic.py:10:1: error[invalid-paramspec] The name of a `ParamSpec` (`NotIt`) must match the name of the variable it is assigned to (`WrongName`)
generics_paramspec_basic.py:23:20: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `P@func1`
generics_paramspec_basic.py:27:38: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `int`
@@ -1001,5 +1003,5 @@
typeddicts_usage.py:28:17: error[missing-typed-dict-key] Missing required key 'name' in TypedDict `Movie` constructor
typeddicts_usage.py:28:18: error[invalid-key] Unknown key "title" for TypedDict `Movie`: Unknown key "title"
typeddicts_usage.py:40:24: error[invalid-type-form] The special form `typing.TypedDict` is not allowed in type expressions. Did you mean to use a concrete TypedDict or `collections.abc.Mapping[str, object]` instead?
-Found 1003 diagnostics
+Found 1005 diagnostics
WARN A fatal error occurred while checking some files. Not all project files were analyzed. See the diagnostics list above for details.
|
Contributor
|
The implicit specialization should use the To be honest I'm not sure if we want So I suspect the better option here might be to change that test assertion, with a comment saying something like " And then add another test asserting explicitly that I think the line you pointed to in |
oconnor663
commented
Nov 21, 2025
| static_assert(is_assignable_to(TypeOf[Bar[bool]], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar[Any]], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar[Unknown]], type[Foo[int]])) |
Contributor
Author
There was a problem hiding this comment.
Is this new Unknown case redundant with the Any case right above it, or are is their behavior different enough that it makes sense to test both?
Contributor
There was a problem hiding this comment.
It's redundant, we very rarely handle Unknown differently from Any. But there's also no harm in including it, don't really care either way.
carljm
approved these changes
Nov 21, 2025
| static_assert(is_assignable_to(TypeOf[Bar[bool]], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar[Any]], type[Foo[int]])) | ||
| static_assert(is_assignable_to(TypeOf[Bar[Unknown]], type[Foo[int]])) |
Contributor
There was a problem hiding this comment.
It's redundant, we very rarely handle Unknown differently from Any. But there's also no harm in including it, don't really care either way.
Member
Looks like we're panicking on a mypy_primer project in CI? |
Contributor
|
Nice catch :) Ok then, don't ship it quite yet... |
Contributor
Author
|
What's one little panic between friends? |
Contributor
Author
|
Progress, this seems to be a minimal repro: class Foo:
pass
_: type[Foo[int]]I think this is the panic that we get when something is inferred more than once. Did it used to have a better error message than this? |
carljm
reviewed
Nov 21, 2025
| } | ||
| None => { | ||
| // TODO: emit a diagnostic if you try to specialize a non-generic class. | ||
| self.infer_type_expression(slice); |
Contributor
There was a problem hiding this comment.
This looks like the likely cause of the panic, slice includes value which we already inferred above.
Contributor
Author
There was a problem hiding this comment.
Yep already there.
Contributor
Author
There was a problem hiding this comment.
Interesting that the line in the wild that triggers this is https://github.com/python-jsonschema/referencing/blob/d97e65e86c95d51f01e18d89461da6d4a0c4e72a/referencing/_core.py#L200
default_specification: (
type[Specification[D]] | Specification[D]
) = Specification,But in this context, Specification is generic...
Contributor
There was a problem hiding this comment.
Oh, that's strange and deserving of further investigation?
But doesn't need to block landing this IMO.
|
Contributor
Author
|
CI is green. Looking at some of the new diagnostics. For example in parameter_storage_class: type[MultiDict[str, t.Any]] = ImmutableMultiDictIt seems like that's a true positive, and that the annotation should be |
Contributor
|
Hmmm, no, I think those ecosystem results might suggest we need to make an adjustment here... and I might have been wrong on the original question about the failing test line. I think we need to default-specialize in that assignment and consider the class object What's not totally clear to me is at which level we should do that default specialization. Should we do it on the name load? At the assignability check? |
Contributor
|
The conformance suite results are showing a similar problem. I think the use of |
Contributor
Author
|
At the risk of biting off more than I can chew, do we want to revisit this: // A non-generic class is never equivalent to a generic class.
// Two non-generic classes are only equivalent if they are equal (handled above).
(ClassType::NonGeneric(_), _) | (_, ClassType::NonGeneric(_)) => {
ConstraintSet::from(false)
}I'll try it just for kicks... |
Contributor
|
No, I don't think that's the right location for the fix. I think the problem is that we sometimes use We have I think maybe the narrow version of the fix here would be located at https://github.com/astral-sh/ruff/blob/main/crates/ty_python_semantic/src/types.rs#L2381 -- that is at least one place where we are clearly constructing a This fix is essentially saying that we consider I suspect that fix may be enough to get the werkzeug example passing, and (depending what the rest of the ecosystem looks like) may be enough to make this landable. The next follow-up (which could be interesting to try here, in case it just works smoothly, but if it reveals more hard problems may be better deferred) would be to actually add some assertions in The next thing which we'd want to look at soon is fixing those conformance suite results. I think we may want to actually make an upstream pull request to the conformance suite to adjust those tests, because I think the way they are written is over-specified. They are asserting that the expression |
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.
Closes astral-sh/ty#1101.
As I push this PR, there is one failing test line on my box, which I need help interpreting. Here's an extracted version of it:
That
static_assertfails. The "never" result seems to originate on this line, which I read as "no non-generic class literal is ever assignable/etc to a specialization" (not sure how to word that correctly). That...seems not true...?...so I'm wondering I've somehow skipped the step where the non-generic class literal, which does in fact have generic parameters, gets them implicitly specialized toobject? I'm betting this will be super obvious to someone :)