[ty] Track argument variance for literal promotion without relying on `SpecializationBuilder` internals by dcreager · Pull Request #21905 · astral-sh/ruff

dcreager · 2025-12-10T20:05:39Z

When inferring a specialization for a generic function call, we might want to promote Literal types that appear in the inferred specialization. Whether we do so depends (among other things) on the variance of those typevars in the parameter types that each argument is matched to. That means we need to track the variance of each typevar as we process the arguments during inference.

Before, we were piggy-backing on how SpecializationBuilder walks through the formal and actual types. Its infer_map method takes in a callback that is invoked each time we record a new type mapping for a typevar. Before, this method was provided with the variance of that typevar, according to the formal/actual pair that we just checked.

We are in the process of changing the internals of SpecializationBuilder so that this approach is no longer tenable. This PR updates the call inference logic to calculate the typevar variance separately, without making any assumptions about how SpecializationBuilder does its work.

astral-sh-bot · 2025-12-10T20:07:42Z

Diagnostic diff on typing conformance tests

No changes detected when running ty on typing conformance tests ✅

astral-sh-bot · 2025-12-10T20:09:39Z

`mypy_primer` results

Changes were detected when running on open source projects

beartype (https://github.com/beartype/beartype)
- beartype/claw/_package/clawpkgtrie.py:66:29: warning[unsupported-base] Unsupported class base with type `<class 'dict[str, PackagesTrieBlacklist]'> | <class 'dict[str, Divergent]'>`
- beartype/claw/_package/clawpkgtrie.py:247:29: warning[unsupported-base] Unsupported class base with type `<class 'dict[str, PackagesTrieWhitelist]'> | <class 'dict[str, Divergent]'>`
- Found 491 diagnostics
+ Found 489 diagnostics

scikit-build-core (https://github.com/scikit-build/scikit-build-core)
- src/scikit_build_core/build/wheel.py:98:20: error[no-matching-overload] No overload of bound method `__init__` matches arguments
- Found 43 diagnostics
+ Found 42 diagnostics

bokeh (https://github.com/bokeh/bokeh)
- src/bokeh/models/annotations/geometry.py:222:29: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[int]] | Property[int]`, found `<class 'Float'>`
+ src/bokeh/models/annotations/geometry.py:222:29: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Float'>`
- src/bokeh/models/annotations/geometry.py:229:30: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[int]] | Property[int]`, found `<class 'Float'>`
+ src/bokeh/models/annotations/geometry.py:229:30: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Float'>`
+ src/bokeh/models/plots.py:779:27: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[5]]] | Property[Literal[5]]`, found `<class 'Int'>`
+ src/bokeh/models/plots.py:801:30: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[2000]]] | Property[Literal[2000]]`, found `<class 'Int'>`
- src/bokeh/models/ranges.py:403:22: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[int]] | Property[int]`, found `<class 'Float'>`
+ src/bokeh/models/ranges.py:403:22: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Float'>`
- src/bokeh/models/ranges.py:413:20: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[int]] | Property[int]`, found `<class 'Float'>`
+ src/bokeh/models/ranges.py:413:20: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Float'>`
+ src/bokeh/models/tools.py:685:25: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Int'>`
+ src/bokeh/models/tools.py:1182:25: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Int'>`
+ src/bokeh/models/widgets/inputs.py:441:34: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[2]]] | Property[Literal[2]]`, found `<class 'Int'>`
+ src/bokeh/models/widgets/inputs.py:592:32: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[100]]] | Property[Literal[100]]`, found `<class 'Int'>`
+ src/bokeh/models/widgets/inputs.py:601:22: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[1]]] | Property[Literal[1]]`, found `<class 'Int'>`
+ src/bokeh/models/widgets/sliders.py:102:22: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[""]]] | Property[Literal[""]]`, found `<class 'String'>`
+ src/bokeh/models/widgets/tables.py:848:31: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `type[Property[Literal[0]]] | Property[Literal[0]]`, found `<class 'Int'>`
- src/bokeh/plotting/_figure.py:201:97: error[invalid-argument-type] Argument to function `process_axis_and_grid` is incorrect: Expected `str | BaseText | None`, found `Unknown | Nullable[str]`
+ src/bokeh/plotting/_figure.py:201:97: error[invalid-argument-type] Argument to function `process_axis_and_grid` is incorrect: Expected `str | BaseText | None`, found `Unknown | Nullable[Literal[""]]`
- src/bokeh/plotting/_figure.py:202:97: error[invalid-argument-type] Argument to function `process_axis_and_grid` is incorrect: Expected `str | BaseText | None`, found `Unknown | Nullable[str]`
+ src/bokeh/plotting/_figure.py:202:97: error[invalid-argument-type] Argument to function `process_axis_and_grid` is incorrect: Expected `str | BaseText | None`, found `Unknown | Nullable[Literal[""]]`
- src/bokeh/settings.py:747:46: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `((int | None | str, /) -> int | None | str) | None`, found `def convert_logging(value: str | int) -> int | None`
+ src/bokeh/settings.py:747:46: error[invalid-argument-type] Argument to bound method `__init__` is incorrect: Expected `((int | None | str, /) -> int | None | Literal["none", "debug"]) | None`, found `def convert_logging(value: str | int) -> int | None`
- Found 867 diagnostics
+ Found 876 diagnostics

zulip (https://github.com/zulip/zulip)
+ zerver/tests/test_users.py:259:25: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["email"], Unknown].__getitem__(key: Literal["email"], /) -> Unknown` cannot be called with key of type `Literal["is_owner"]` on object of type `dict[Literal["email"], Unknown]`
+ zerver/tests/test_users.py:261:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["email"], Unknown].__getitem__(key: Literal["email"], /) -> Unknown` cannot be called with key of type `Literal["is_owner"]` on object of type `dict[Literal["email"], Unknown]`
+ zerver/tests/test_users.py:318:25: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["email"], Unknown].__getitem__(key: Literal["email"], /) -> Unknown` cannot be called with key of type `Literal["is_admin"]` on object of type `dict[Literal["email"], Unknown]`
+ zerver/tests/test_users.py:320:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["email"], Unknown].__getitem__(key: Literal["email"], /) -> Unknown` cannot be called with key of type `Literal["is_admin"]` on object of type `dict[Literal["email"], Unknown]`
+ zerver/tests/test_users.py:361:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["email"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:362:27: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["avatar_url"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:363:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["delivery_email"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:387:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["email"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:389:13: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["avatar_url"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:402:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["email"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:404:13: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["avatar_url"]` on object of type `dict[Literal["user_id"], Unknown]`
+ zerver/tests/test_users.py:406:26: error[invalid-argument-type] Method `__getitem__` of type `bound method dict[Literal["user_id"], Unknown].__getitem__(key: Literal["user_id"], /) -> Unknown` cannot be called with key of type `Literal["delivery_email"]` on object of type `dict[Literal["user_id"], Unknown]`
- Found 3631 diagnostics
+ Found 3643 diagnostics

No memory usage changes detected ✅

carljm · 2025-12-10T20:23:51Z

Code looks fine. Ecosystem changes on zulip suggest that there's an unintentional behavior change here?

astral-sh-bot · 2025-12-10T20:32:43Z

`ecosystem-analyzer` results

Lint rule	Added	Removed	Changed
`invalid-argument-type`	21	0	7
`unused-ignore-comment`	0	3	0
Total	21	3	7

Full report with detailed diff (timing results)

carljm · 2025-12-10T20:58:13Z

It seems like in the zulip cases we are failing to do literal promotion in a case where previously we did promote.

dcreager · 2025-12-11T16:32:55Z

It seems like in the zulip cases we are failing to do literal promotion in a case where previously we did promote.

Fixed the zulip case and added an mdtest regression for it.

codspeed-hq · 2025-12-11T16:44:52Z

CodSpeed Performance Report

Merging #21905 will degrade performances by 4.25%

_{Comparing dcreager/param-variance (fe25ee2) with main (c8851ec)}

Summary

❌ 1 regression
✅ 12 untouched
⏩ 39 skipped¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Mode	Benchmark	`BASE`	`HEAD`	Change
❌	Simulation	`hydra-zen`	1.3 s	1.4 s	-4.25%

39 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

AlexWaygood · 2025-12-11T18:05:16Z

Looks like both the ecosystem CI jobs timed out, which I think is probably not related to the github outage?

dcreager · 2025-12-11T18:08:38Z

Looks like both the ecosystem CI jobs timed out, which I think is probably not related to the github outage?

Yep, I've got a mypy_primer run going locally to confirm

* origin/main: (29 commits) Document range suppressions, reorganize suppression docs (#21884) Ignore ruff:isort like ruff:noqa in new suppressions (#21922) [ty] Handle `Definition`s in `SemanticModel::scope` (#21919) [ty] Attach salsa db when running ide tests for easier debugging (#21917) [ty] Don't show hover for expressions with no inferred type (#21924) [ty] avoid unions of generic aliases of the same class in fixpoint (#21909) [ty] Squash false positive logs for failing to find `builtins` as a real module [ty] Uniformly use "not supported" in diagnostics (#21916) [ty] Reduce size of ty-ide snapshots (#21915) [ty] Adjust scope completions to use all reachable symbols [ty] Rename `all_members_of_scope` to `all_end_of_scope_members` [ty] Remove `all_` prefix from some routines on UseDefMap Enable `--document-private-items` for `ruff_python_formatter` (#21903) Remove `BackwardsTokenizer` based `parenthesized_range` references in `ruff_linter` (#21836) [ty] Revert "Do not infer types for invalid binary expressions in annotations" (#21914) Skip over trivia tokens after re-lexing (#21895) [ty] Avoid inferring types for invalid binary expressions in string annotations (#21911) [ty] Improve overload call resolution tracing (#21913) [ty] fix missing heap_size on Salsa query (#21912) [ty] Support implicit type of `cls` in signatures (#21771) ...

dcreager · 2025-12-11T20:26:12Z

Looks like both the ecosystem CI jobs timed out

Cherry-picked several of performance optimizations commits from #21551 and that seems to have fixed the timeouts. Locally, at least! Waiting for CI to confirm

dcreager · 2025-12-11T21:12:48Z

Locally, at least! Waiting for CI to confirm

Welp, famous last words! Instead of pushing this forward more, I'm going to see if I can update the generic protocol PR to not require this.

dcreager · 2025-12-15T20:56:47Z

I'm going to see if I can update the generic protocol PR to not require this.

This happened, so I'm closing this for now

dcreager added 5 commits December 10, 2025 13:59

a good start

e2c9a07

don't track self

3572c85

stop tracking variance

8d57489

clean this up a bit

095c2bb

more human comment

8ec6673

dcreager requested review from AlexWaygood, carljm and sharkdp as code owners December 10, 2025 20:05

dcreager added internal An internal refactor or improvement ty Multi-file analysis & type inference labels Dec 10, 2025

carljm added the ecosystem-analyzer label Dec 10, 2025

dcreager added 3 commits December 11, 2025 11:15

add ConstraintSetAssignability relation

b9691b8

add failing test

745f4d9

better predicate for skipping typevars

dbab2c4

dcreager added ecosystem-analyzer and removed ecosystem-analyzer labels Dec 11, 2025

dcreager added 5 commits December 11, 2025 14:36

don't create T ≤ T constraints

312c4ce

use ConstraintSetAssignability for constraint bounds

db3501b

don't pivot on never or object

7fe09c2

group typevars by binding context

3cb4b0e

abstract over any mention of a typevar

438de79

dcreager force-pushed the dcreager/param-variance branch from df40a2b to 438de79 Compare December 11, 2025 20:20

format

fe25ee2

dcreager closed this Dec 15, 2025

dcreager deleted the dcreager/param-variance branch December 15, 2025 20:56

ibraheemdev mentioned this pull request Jan 1, 2026

[ty] Inform literal promotion with variance of inferred argument types #22322

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[ty] Track argument variance for literal promotion without relying on `SpecializationBuilder` internals#21905

[ty] Track argument variance for literal promotion without relying on `SpecializationBuilder` internals#21905
dcreager wants to merge 15 commits intomainfrom
dcreager/param-variance

dcreager commented Dec 10, 2025

Uh oh!

astral-sh-bot bot commented Dec 10, 2025 •

edited

Loading

Uh oh!

astral-sh-bot bot commented Dec 10, 2025

Uh oh!

carljm commented Dec 10, 2025

Uh oh!

astral-sh-bot bot commented Dec 10, 2025

Uh oh!

carljm commented Dec 10, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

codspeed-hq bot commented Dec 11, 2025 •

edited

Loading

Uh oh!

AlexWaygood commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 15, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

dcreager commented Dec 10, 2025

Uh oh!

astral-sh-bot bot commented Dec 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Diagnostic diff on typing conformance tests

Uh oh!

astral-sh-bot bot commented Dec 10, 2025

mypy_primer results

Uh oh!

carljm commented Dec 10, 2025

Uh oh!

astral-sh-bot bot commented Dec 10, 2025

ecosystem-analyzer results

Uh oh!

carljm commented Dec 10, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

codspeed-hq bot commented Dec 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CodSpeed Performance Report

Merging #21905 will degrade performances by 4.25%

Summary

Benchmarks breakdown

Footnotes

Uh oh!

AlexWaygood commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 11, 2025

Uh oh!

dcreager commented Dec 15, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

astral-sh-bot bot commented Dec 10, 2025 •

edited

Loading

`mypy_primer` results

`ecosystem-analyzer` results

codspeed-hq bot commented Dec 11, 2025 •

edited

Loading