Skip to content

Add RegularCallableTypeOf and into_regular_callable in ty_extensions - #23909

Merged
dhruvmanila merged 8 commits into
mainfrom
dhruv/callable-type-of-regression
Mar 13, 2026
Merged

dhruvmanila merged 8 commits into
mainfrom
dhruv/callable-type-of-regression

Conversation

@dhruvmanila

Copy link
Copy Markdown
Member

Summary

fixes: astral-sh/ty#3020

Test Plan

Add a regression test case that currently fails on main.

@dhruvmanila dhruvmanila added the testing Related to testing Ruff or ty label Mar 12, 2026
@dhruvmanila dhruvmanila added the ty The ty type checker label Mar 12, 2026
@astral-sh-bot
astral-sh-bot Bot requested a review from oconnor663 March 12, 2026 05:31
@astral-sh-bot

astral-sh-bot Bot commented Mar 12, 2026 •

Copy link
Copy Markdown

Typing conformance results

No changes detected ✅

Current numbers
The percentage of diagnostics emitted that were expected errors held steady at 85.29%. The percentage of expected errors that received a diagnostic held steady at 78.13%. The number of fully passing files held steady at 64/132.

@astral-sh-bot

astral-sh-bot Bot commented Mar 12, 2026 •

Copy link
Copy Markdown

mypy_primer results

Changes were detected when running on open source projects
scikit-build-core (https://github.com/scikit-build/scikit-build-core)
+ src/scikit_build_core/build/wheel.py:99:20: error[no-matching-overload] No overload of bound method `__init__` matches arguments
- Found 59 diagnostics
+ Found 60 diagnostics

@astral-sh-bot

astral-sh-bot Bot commented Mar 12, 2026 •

Copy link
Copy Markdown

Memory usage report

Memory usage unchanged ✅

@dhruvmanila

Copy link
Copy Markdown
Member Author

Oh, of course this would mean that the test cases that is relying on this behavior would fail 🤦

@dhruvmanila
dhruvmanila marked this pull request as draft March 12, 2026 05:41
@dhruvmanila dhruvmanila changed the title Update CallableTypeOf to always use Regular callable type kind Add RegularCallableTypeOf and into_regular_type in ty_extensions Mar 12, 2026
@dhruvmanila

Copy link
Copy Markdown
Member Author

There are some things to do:

  • Change all existing references to use these "regular" variants except for the few that actually requires using the existing variants
  • Add documentation to explain the difference between these two variants

But, I'm opening this up for feedback (and name bike shedding ;)) before I make those large changes.

@dhruvmanila
dhruvmanila marked this pull request as ready for review March 12, 2026 11:48
Comment on lines +16 to +21
def f(a: int, b: str, /) -> None: ...

static_assert(is_assignable_to(Callable[[int, str], None], RegularCallableTypeOf[f]))
static_assert(is_subtype_of(Callable[[int, str], None], RegularCallableTypeOf[f]))
static_assert(is_assignable_to(Callable[[int, str], None], TypeOf[into_regular_callable(f)]))
static_assert(is_subtype_of(Callable[[int, str], None], TypeOf[into_regular_callable(f)]))

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I guess another solution would be to use a callback protocol for these kinds of assertions? This protocol:

class Foo(Protocol):
    def __call__(self, *, x: int): ...

is equivalent to this type:

def func(*, x: int): ...

type Foo = RegularCallableTypeOf[func]

and then we don't need another special form. But the disadvantage of that is that we're relying on our Protocol subtyping/assignability checks being correct, and there are some complications there. So it's probably better to go with what you have right now.

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, definitely! I think on a long-term horizon Protocol is the way to go here and then we can remove these variants.

Relatedly, I've been using Protocol to do these relation checks locally sometimes and I haven't faced any issues yet :)

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@AlexWaygood AlexWaygood changed the title Add RegularCallableTypeOf and into_regular_type in ty_extensions Add RegularCallableTypeOf and into_regular_callable in ty_extensions Mar 12, 2026
@charliermarsh

Copy link
Copy Markdown
Member

Superficially this reminds me a bit of astral-sh/ty#1452 / #23560

@dhruvmanila
dhruvmanila force-pushed the dhruv/callable-type-of-regression branch from 4c85833 to 7592954 Compare March 12, 2026 16:04
@dhruvmanila

Copy link
Copy Markdown
Member Author

(Going to look at the replacement again tomorrow morning to make sure they're all correct and then merge.)

@carljm
carljm removed their request for review March 13, 2026 01:18
@dhruvmanila

Copy link
Copy Markdown
Member Author

So, the convention is to always use the regular variants unless you explicitly wants to test against the function-like behavior of the callable.

Another convention would've been to always use the regular variants when you're checking a relation between two callables except in cases where you want to check the relation which involves at least one function-like callable, and use non-regular variant in other tests.

I'm going with the former in this PR.

@dhruvmanila
dhruvmanila merged commit ce8f75f into main Mar 13, 2026
51 checks passed
@dhruvmanila
dhruvmanila deleted the dhruv/callable-type-of-regression branch March 13, 2026 03:56
carljm added a commit that referenced this pull request Mar 13, 2026
* main: (94 commits)
  Fix shell injection via `shell=True` in subprocess calls (#23894)
  [ty] Refactor `relation.rs` to store state on a struct rather than passing around 7 arguments every time we recurse (#23837)
  Don't return code actions for non-Python documents (#23905)
  [ty] Make the default database truly statically infallible (#23929)
  [ty] Add `Download` button to ty playground which creates a zip export (#23478)
  [ty] Respect `kw_only` overwrites in dataclasses (#23930)
  [ty] Clarify in diagnostics that `from __future__ import annotations` only stringifies type annotations (#23928)
  [ty]  Add a `Copy Markdown` button to playground (#23002)
  [ty] Fix folding range classification of lines starting with `#` (#23831)
  [ty] Fix folding ranges for notebooks (#23830)
  [ty] fix too-many-cycle panics when inferring literal type loop variables (#23875)
  Add `RegularCallableTypeOf` and `into_regular_callable` in `ty_extensions` (#23909)
  [ty] treat properties as full structural types (#23925)
  [ty] Avoid duplicated work during multi-inference (#23923)
  [ty]: make `possibly-missing-attribute` ignored by default
  [ty]: split out `possibly-missing-submodule` from `possibly-missing-attribute`
  Update astral-sh/setup-uv action to v7.5.0 (#23922)
  [ty] Show truthiness in ConstraintSet display and simplify falsy error message (#23913)
  Bump 0.15.6 (#23919)
  [ty] Narrow type context during collection literal inference (#23844)
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

testing Related to testing Ruff or ty ty The ty type checker

Projects

None yet

Development

Successfully merging this pull request may close these issues.

ty_extensions.CallableTypeOf always considers the callable as function-like

4 participants