[ty] Fix overloaded callable assignability for unary Callable targets#23277
Merged
dcreager merged 7 commits intoastral-sh:mainfrom Feb 26, 2026
Merged
Conversation
Typing conformance results improved 🎉The percentage of diagnostics emitted that were expected errors increased from 84.99% to 85.17%. The percentage of expected errors that received a diagnostic held steady at 75.25%. Summary
False positives removedDetails
|
Memory usage reportMemory usage unchanged ✅ |
Merging this PR will not alter performance
Comparing 
Footnotes
|
|
4cbba88 to
a018f28
Compare
a018f28 to
85da64d
Compare
Contributor
Author
|
primer results look good overall. From my review, the remaining diffs are either intended behavior changes or primer flakiness and I couldn’t reproduce a stable new regression attributable to this PR. (I did find real regressions earlier in development but those have been addressed) |
|
| Lint rule | Added | Removed | Changed |
|---|---|---|---|
invalid-argument-type |
0 | 3 | 0 |
invalid-method-override |
0 | 3 | 0 |
unresolved-attribute |
1 | 0 | 0 |
| Total | 1 | 6 | 0 |
Contributor
|
@dcreager It would be great if you could review this. I'd love to continue to work on dataclass converters to improve our typing conformance results in that area, but this issue is blocking progress. Not very high priority, though. |
dcreager
reviewed
Feb 26, 2026
crates/ty_python_semantic/resources/mdtest/generics/legacy/callables.md
Outdated
Show resolved
Hide resolved
Member
There was a problem hiding this comment.
It can be onerous, but I prefer for all of the new tests in legacy to have analogues in pep695. We've definitely encountered subtle bugs that stem from how we treat those separately in some cases.
Member
There was a problem hiding this comment.
Done. I copied the generics-related tests into pep695 to test the other typevar syntax. The first few tests were not related to generics at all, so I moved those into annotations/callable.md
| relation: TypeRelation<'db>, | ||
| ) -> Option<ConstraintSet<'db>> { | ||
| let other_parameter_type = | ||
| Self::single_required_positional_parameter_type(other_signature)?; |
Member
There was a problem hiding this comment.
Given this usage, I can see why you put these methods here instead of on the types that they take in. If you want to emphasize that those are helpers that are only needed for this method, an alternative formulation that we've used is to declare them as closures inside this method body:
let single_require_positional_parameter_type = |signature| {
...
};
let is_unary_overload_aggregate_candidate_type = |ty| {
// note that you probably won't need `db` as a param since a
// closure can access its surrounding scope
...
};
Contributor
Author
Sorry I couldn't get to it (currently at work), I saw you handled everything! Thanks a lot for the review and for your work :) |
dcreager
approved these changes
Feb 26, 2026
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
4 participants
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
Fixes astral-sh/ty#2546
Improves assignability checking from overloaded callables to a single
Callable[...]target with an explicit union parameter domain.Previously, this was handled with per-overload
when_anychecks. This PR replaces that with an aggregate probe over the overload set that:The aggregate probe is accept-only ie. if it isn't definitively satisfied, we fall back to the existing
when_anybehavior. This change is intentionally scoped to unary targets with explicit union domains and excludes dynamic/typevar candidates. Generaln > 1overload-set assignability is a way larger problem left for later.Test Plan
Callable[[T], T]under union-constrained inference.overloaded_converter,ConverterClass) with an explicit TODO expectation for the still unhandledconverter=dictcase.