[ty] Promote literals in invariant return position #21320
Closed
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
Contributor
Diagnostic diff on typing conformance testsChanges were detected when running ty on typing conformance tests--- old-output.txt 2025-11-10 23:05:16.349873661 +0000
+++ new-output.txt 2025-11-10 23:05:19.738890404 +0000
@@ -688,6 +688,8 @@
literals_interactions.py:60:49: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `Matrix[A@Matrix, B@Matrix]`
literals_interactions.py:63:52: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `Matrix[A@Matrix, C@__matmul__]`
literals_interactions.py:66:28: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `Matrix[B@Matrix, A@Matrix]`
+literals_interactions.py:71:9: error[unsupported-operator] Operator `@` is unsupported between objects of type `Matrix[Literal[2], Literal[3]]` and `Matrix[Literal[3], Literal[7]]`
+literals_interactions.py:72:5: error[type-assertion-failure] Argument does not have asserted type `Matrix[Literal[2], Literal[7]]`
literals_interactions.py:106:35: error[invalid-argument-type] Argument to function `expects_bad_status` is incorrect: Expected `Literal["MALFORMED", "ABORTED"]`, found `str`
literals_interactions.py:109:32: error[invalid-argument-type] Argument to function `expects_pending_status` is incorrect: Expected `Literal["PENDING"]`, found `str`
literals_literalstring.py:23:51: error[invalid-return-type] Function always implicitly returns `None`, which is not assignable to return type `LiteralString`
@@ -1007,5 +1009,5 @@
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] Invalid key 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 1009 diagnostics
+Found 1011 diagnostics
WARN A fatal error occurred while checking some files. Not all project files were analyzed. See the diagnostics list above for details.
|
ibraheemdev
force-pushed
the
ibraheem/generic-promotion
branch
from
0888d4b to
0f07e50
Compare
Contributor
|
Member
|
The primer report looks excellent but it looks like there's a new type-assertion failure in the conformance test suite -- have you checked out if that's expected? |
Contributor
Note that both the primer and the conformance checks here include changes from the previous PRs on which this is stacked. We compare against main. |
AlexWaygood
reviewed
Nov 10, 2025
crates/ty_python_semantic/resources/mdtest/generics/pep695/classes.md
Outdated
Show resolved
Hide resolved
crates/ty_python_semantic/resources/mdtest/generics/pep695/classes.md
Outdated
Show resolved
Hide resolved
Comment on lines
+517
to
+518
| reveal_type(frozenset((1, 2, 3))) # revealed: frozenset[Literal[1, 2, 3]] | ||
| reveal_type(frozenset(((1, 2, 3),))) # revealed: frozenset[tuple[Literal[1], Literal[2], Literal[3]]] |
| let promote_literals = |typevar: GenericContextTypeVar<'db>, ty: Type<'db>| -> Type<'db> { | ||
| let bound_typevar = typevar.bound_typevar(); | ||
|
|
||
| if typevar.is_inherited() && bound_typevar.variance(self.db).is_invariant() { |
Member
There was a problem hiding this comment.
What does is_inherited mean here? Why does it matter whether a typevar is inherited or not?
(It would be great if is_inherited could have a doc-comment explaining what it does 😃)
Member
There was a problem hiding this comment.
Ah, I think I see -- you're using the term to mean that the typevar is associated with a class's generic context rather than the generic context of a function or method? So it doesn't have any thing to do with the generic context of a class's base classes etc.?
Member
Author
There was a problem hiding this comment.
That's right, though I didn't invent that terminology.
Ideally we wouldn't need to special case inherited type variables, but we don't currently have a robust way to get access to the bound type of a constructor method (the synthesized_constructor_return_ty).
ibraheemdev
force-pushed
the
ibraheem/generic-constructor-tcx
branch
from
37ac0b5 to
25ed082
Compare
ibraheemdev
force-pushed
the
ibraheem/generic-promotion
branch
from
0f07e50 to
47cdf49
Compare
ibraheemdev
force-pushed
the
ibraheem/generic-promotion
branch
from
4d88df9 to
4da67a1
Compare
Member
Author
|
Hmm this is more tricky than I realized. We can only perform the literal promotion of a given type variable if it does not lead to argument assignability errors. We may need to resort to a simple heuristic of "promote every type variable in invariant position", and fallback to not performing any promotion if type-checking fails, as I suspect attempting partial promotion would be too expensive. |
Member
Author
|
Superseded by #21439. |
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.
Summary
We currently unconditionally promote literals in generic class constructors. We should do this for any literal in an invariant position in the return type. Resolves astral-sh/ty#1357.
This PR is stacked on #20933.