[ty] Fix assignability of intersections with bounded typevars#24502
Merged
AlexWaygood merged 6 commits intomainfrom Apr 9, 2026
Merged
[ty] Fix assignability of intersections with bounded typevars#24502AlexWaygood merged 6 commits intomainfrom
AlexWaygood merged 6 commits intomainfrom
Conversation
Typing conformance resultsNo changes detected ✅Current numbersThe percentage of diagnostics emitted that were expected errors held steady at 87.72%. The percentage of expected errors that received a diagnostic held steady at 82.85%. The number of fully passing files held steady at 74/132. |
Memory usage reportMemory usage unchanged ✅ |
|
| Lint rule | Added | Removed | Changed |
|---|---|---|---|
invalid-await |
0 | 40 | 0 |
invalid-argument-type |
0 | 13 | 0 |
invalid-assignment |
1 | 0 | 0 |
invalid-return-type |
0 | 1 | 0 |
no-matching-overload |
0 | 1 | 0 |
unsupported-operator |
0 | 1 | 0 |
| Total | 1 | 56 | 0 |
Large timing changes:
| Project | Old Time | New Time | Change |
|---|---|---|---|
freqtrade |
0.55s | 0.91s | +65% |
Changes in flaky projects detected. Raw diff output excludes flaky projects; see the HTML report for details.
Raw diff (16 changes)
Tanjun (https://github.com/FasterSpeeding/Tanjun)
- tanjun/annotations.py:1949:74 error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int | float`, found `_NumberT@__getitem__ & ~int`
- tanjun/annotations.py:1996:74 error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int | float`, found `_NumberT@__getitem__ & ~int`
- tanjun/annotations.py:2110:75 error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int | float`, found `_NumberT@__getitem__ & ~float`
- tanjun/annotations.py:2110:86 error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `int | float`, found `_NumberT@__getitem__ & ~float`
- tanjun/annotations.py:2724:26 error[invalid-argument-type] Argument to function `parse_annotated_args` is incorrect: Expected `SlashCommand[Any] | MessageCommand[Any]`, found `_CommandUnionT@with_annotated_args & ~AlwaysFalsy`
altair (https://github.com/vega/altair)
- altair/vegalite/v6/api.py:1179:45 error[invalid-argument-type] Argument to function `_reveal_parsed_shorthand` is incorrect: Expected `Mapping[str, Any]`, found `_C@Then & Top[dict[Unknown, Unknown]]`
artigraph (https://github.com/artigraph/artigraph)
- src/arti/internal/type_hints.py:177:37 error[invalid-argument-type] Argument to function `_check_issubclass` is incorrect: Expected `type`, found `T@lenient_issubclass & ~tuple[object, ...]`
beartype (https://github.com/beartype/beartype)
- beartype/_decor/_nontype/decornontype.py:161:43 error[invalid-argument-type] Argument is incorrect: Argument type `BeartypeableT@beartype_nontype & ~type` does not satisfy upper bound `((...) -> Any) | classmethod[Unknown, (...), Unknown] | property` of type variable `BeartypeableT`
- beartype/_decor/_nontype/decornontype.py:220:47 error[invalid-argument-type] Argument is incorrect: Argument type `BeartypeableT@beartype_nontype & ~type` does not satisfy upper bound `((...) -> Any) | classmethod[Unknown, (...), Unknown] | property` of type variable `BeartypeableT`
- beartype/_decor/decorcore.py:133:26 error[invalid-argument-type] Argument to function `beartype_nontype` is incorrect: Argument type `BeartypeableT@_beartype_object_fatal & ~type` does not satisfy upper bound `((...) -> Any) | classmethod[Unknown, (...), Unknown] | property` of type variable `BeartypeableT`
hydpy (https://github.com/hydpy-dev/hydpy)
- hydpy/auxs/iuhtools.py:436:25 error[unsupported-operator] Unary operator `-` is not supported for object of type `VectorFloatFlex@__call__ | Unknown`
- hydpy/auxs/iuhtools.py:431:17 error[no-matching-overload] No overload of function `clip` matches arguments
+ hydpy/auxs/iuhtools.py:431:17 error[invalid-assignment] Object of type `ndarray[tuple[Any, ...], dtype[Any]]` is not assignable to `VectorFloatFlex@__call__`
setuptools (https://github.com/pypa/setuptools)
- setuptools/glob.py:75:42 error[invalid-argument-type] Argument to function `has_magic` is incorrect: Expected `str | bytes`, found `AnyStr@_iglob & ~AlwaysFalsy`
xarray (https://github.com/pydata/xarray)
- xarray/core/dtypes.py:92:36 error[invalid-argument-type] Argument to function `datetime_data` is incorrect: Expected `str | _HasDType[dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]]] | dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]] | _HasNumPyDType[dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]]]`, found `T_dtype@maybe_promote & dtype[object]`
- xarray/core/dtypes.py:105:36 error[invalid-argument-type] Argument to function `datetime_data` is incorrect: Expected `str | _HasDType[dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]]] | dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]] | _HasNumPyDType[dtype[datetime64[date | int | None] | timedelta64[timedelta | int | None]]]`, found `T_dtype@maybe_promote & dtype[object]`ed2a945 to
53f89db
Compare
Contributor
|
Hmm, I wasn't prioritizing reviving that PR because I lost confidence that this ad hoc handling wherever we encounter these intersections was the right way to approach the problem at all, vs synthesizing derived typevars with narrowed bounds. I think it's on @sharkdp's monthly plan to look into that approach? But if this is looking pretty good, doesn't hurt to do this and then see whether we can/should replace it later. |
Member
Author
|
Yeah, I definitely agree that there's probably a more general solution that eliminates this whole class of bugs. This wasn't hard to put together at all, though, and gets rid of real false positives for users in the meantime. |
carljm
approved these changes
Apr 9, 2026
carljm
reviewed
Apr 9, 2026
31b43cd to
d37a3cb
Compare
ibraheemdev
pushed a commit
that referenced
this pull request
Apr 15, 2026
Co-authored-by: Carl Meyer <carl@astral.sh>
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
3 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
This is basically the same PR as #24193, but it applies the optimisation I suggested in #24193 (comment) to get rid of the huge performance regression reported by codspeed.
It also adds some more tests, since several branches added in #24193 could be removed without any tests failing, but in one case the ecosystem report uncovered cases where the added branch did indeed get rid of false positives. I added a test case based on that snippet surfaced by the ecosystem report.
Note that the ecosystem report is still showing large increases in timings for some projects. Curiously, though, one of those projects is freqtrade, and our codspeed benchmark that uses freqtrade doesn't show any change at all as a result of this PR. On the one hand, our codspeed benchmark doesn't use multithreading, whereas the version of ty run by ecosystem-analyzer does, so maybe this is telling us something about how the new code added here scales with multithreading? On the other hand, though, the ecosystem-analyzer timings use a single noisy sample rather than an average across several measurements, like codspeed does.
Fixes astral-sh/ty#3137
Test Plan
mdtests
Co-authored-by: Carl Meyer carl@astral.sh