Codestin Search App

charliermarsh · 2026-06-13T02:54:39Z

Summary

This PR infers the types of names introduced by match patterns.

def unpack(value: tuple[int, str]) -> None:
    match value:
        case [first, second] as whole:
            reveal_type(first)   # int
            reveal_type(second)  # str
            reveal_type(whole)   # tuple[int, str]

The implementation analyzes a successful match recursively: each pattern receives the type of the value it is matching, computes the type that can successfully match, and passes the appropriate extracted type to its children. A capture binds the type it receives; an as pattern binds the successful type of its complete nested pattern.

The analysis also distinguishes facts established while a pattern is being evaluated from the stable type assigned to an alias. An alias for a mutable sequence keeps only its sequence type, because later mutation can invalidate the observed shape; exact tuples can retain facts learned from their children. For example, we don't narrow here:

def f(value: list[int | str]) -> None:
    match value:
        case [int(), str()] as whole:
            reveal_type(whole)       # list[int | str]
            reveal_type(len(whole))  # int, not Literal[2]
            reveal_type(whole[0])    # int | str, not int
            reveal_type(whole[1])    # int | str, not str

But we do narrow for an exact tuple:

def f(value: tuple[int | str, int | str]) -> None:
    match value:
        case [int(), str()] as whole:
            reveal_type(len(whole))  # Literal[2]
            reveal_type(whole[0])    # int
            reveal_type(whole[1])    # str

astral-sh-bot · 2026-06-13T02:56:23Z

Typing conformance results

The percentage of diagnostics emitted that were expected errors held steady at 94.37%. The percentage of expected errors that received a diagnostic held steady at 89.00%. The number of fully passing files held steady at 94/134.

Summary

How are test cases classified?

Each test case represents one expected error annotation or a group of annotations sharing a tag. Counts are per test case, not per diagnostic — multiple diagnostics on the same line count as one. Required annotations (E) are true positives when ty flags the expected location and false negatives when it does not. Optional annotations (E?) are true positives when flagged but true negatives (not false negatives) when not. Tagged annotations (E[tag]) require ty to flag exactly one of the tagged lines; tagged multi-annotations (E[tag+]) allow any number up to the tag count. Flagging unexpected locations counts as a false positive.

Metric	Old	New	Diff
True Positives	955	955	+0
False Positives	57	57	+0
False Negatives	118	118	+0
Total Diagnostics	1060	1060	+0
Precision	94.37%	94.37%	+0.00%
Recall	89.00%	89.00%	+0.00%
Passing Files	94/134	94/134	+0

True positives changed (2)

2 diagnostics

Test case

Diff

tuples_type_compat.py:106

-error[type-assertion-failure] Type `tuple[str, str] | (tuple[int, *tuple[str, ...], int] & <Protocol with members '__getitem__', '__len__'> & ~<Protocol with members '__getitem__', '__len__'>)` does not match asserted type `tuple[str, str] | tuple[int, int]`
+error[type-assertion-failure] Type `tuple[str, str] | (tuple[int, *tuple[str, ...], int] & <Protocol with members '__getitem__', '__len__'>)` does not match asserted type `tuple[str, str] | tuple[int, int]`

tuples_type_compat.py:111

-error[type-assertion-failure] Type `tuple[int, *tuple[str, ...], int] & <Protocol with members '__getitem__', '__len__'> & ~<Protocol with members '__getitem__', '__len__'> & ~<Protocol with members '__getitem__', '__len__'>` does not match asserted type `tuple[int, str, int]`
+error[type-assertion-failure] Type `tuple[int, *tuple[str, ...], int] & <Protocol with members '__getitem__', '__len__'>` does not match asserted type `tuple[int, str, int]`

astral-sh-bot · 2026-06-13T02:57:08Z

Memory usage report

Memory usage unchanged ✅

astral-sh-bot · 2026-06-13T02:59:01Z

`ecosystem-analyzer` results

Lint rule	Added	Changed
`invalid-argument-type`	13	0
`invalid-assignment`	1	3
`type-assertion-failure`	0	3
`unresolved-attribute`	1	1
`invalid-return-type`	1	0
Total	16	7

Large timing changes:

Project	Old Time	New Time	Change
`egglog-python`	0.20s	0.31s	+54%

Flaky changes detected. This PR summary excludes flaky changes; see the HTML report for details.

Raw diff (23 changes)

cibuildwheel (https://github.com/pypa/cibuildwheel)
+ cibuildwheel/options.py:428:50 error[invalid-argument-type] Argument to bound method `OptionFormat.format_list` is incorrect: Expected `Sequence[str | int]`, found `(Mapping[str, Sequence[str | int] | int] & Sequence[object] & ~str & ~bytes & ~bytearray) | (Sequence[str | int] & ~str & ~bytes & ~bytearray) | (int & Sequence[object])`

cwltool (https://github.com/common-workflow-language/cwltool)
+ cwltool/secrets.py:46:40 error[invalid-argument-type] Argument to bound method `SecretStore.has_secret` is incorrect: Expected `int | str | float | ... omitted 4 union elements`, found `object`
+ cwltool/secrets.py:50:40 error[invalid-argument-type] Argument to bound method `SecretStore.has_secret` is incorrect: Expected `int | str | float | ... omitted 4 union elements`, found `object`
+ cwltool/secrets.py:62:24 error[invalid-return-type] Return type does not match returned value: expected `int | str | float | ... omitted 4 union elements`, found `dict[object, int | str | float | ... omitted 4 union elements]`
+ cwltool/secrets.py:62:42 error[invalid-argument-type] Argument to bound method `SecretStore.retrieve` is incorrect: Expected `int | str | float | ... omitted 4 union elements`, found `object`
+ cwltool/secrets.py:64:39 error[invalid-argument-type] Argument to bound method `SecretStore.retrieve` is incorrect: Expected `int | str | float | ... omitted 4 union elements`, found `object`
+ cwltool/process.py:507:44 error[invalid-argument-type] Argument to function `var_spool_cwl_detector` is incorrect: Expected `int | str | float | ... omitted 5 union elements`, found `object`
+ cwltool/process.py:510:44 error[invalid-argument-type] Argument to function `var_spool_cwl_detector` is incorrect: Expected `int | str | float | ... omitted 5 union elements`, found `object`

egglog-python (https://github.com/egraphs-good/egglog-python)
- python/egglog/egraph_state.py:378:25 error[type-assertion-failure] Type `@Todo & ~Literal["delete"] & ~Literal["subsume"]` is not equivalent to `Never`
+ python/egglog/egraph_state.py:378:25 error[type-assertion-failure] Type `Unknown & ~Literal["delete"] & ~Literal["subsume"]` is not equivalent to `Never`
- python/egglog/egraph_state.py:623:25 error[type-assertion-failure] Type `@Todo & ~None & ~int & ~float & ~str` is not equivalent to `Never`
+ python/egglog/egraph_state.py:623:25 error[type-assertion-failure] Type `Unknown & ~None & ~int & ~float & ~str` is not equivalent to `Never`
- python/egglog/pretty.py:309:17 error[type-assertion-failure] Type `@Todo & ~None & ~int & ~float & ~str` is not equivalent to `Never`
+ python/egglog/pretty.py:309:17 error[type-assertion-failure] Type `Unknown & ~None & ~int & ~float & ~str` is not equivalent to `Never`

jax (https://github.com/google/jax)
+ jax/_src/pallas/mosaic_gpu/lowering.py:1666:55 error[invalid-argument-type] Argument to bound method `UnswizzleRef.commute_reshape` is incorrect: Expected `ShapedArray`, found `AbstractValue`
+ jax/experimental/mosaic/gpu/layout_inference.py:424:10 error[invalid-assignment] Object of type `WGStridedFragLayout & ~WGSplatFragLayout` is not assignable to `RegisterLayout`
+ jax/experimental/mosaic/gpu/layout_inference.py:425:54 error[unresolved-attribute] Object of type `RegisterLayout` has no attribute `vec_size`
+ jax/experimental/mosaic/gpu/layout_inference.py:426:51 error[invalid-argument-type] Argument to function `is_supported_strided_layout_broadcast` is incorrect: Expected `WGStridedFragLayout`, found `RegisterLayout`

kopf (https://github.com/nolar/kopf)
- kopf/_cogs/structs/dicts.py:244:29 error[unresolved-attribute] Object of type `(_T@walk & ~None) | Iterable[_T@walk | Iterable[_T@walk]]` has no attribute `obj`
+ kopf/_cogs/structs/dicts.py:244:29 error[unresolved-attribute] Attribute `obj` is not defined on `_T@walk & _dummy`, `Iterable[_T@walk | Iterable[_T@walk]] & _dummy` in union `(_T@walk & Unknown & ~None) | (Iterable[_T@walk | Iterable[_T@walk]] & Unknown) | (_T@walk & _dummy) | (Iterable[_T@walk | Iterable[_T@walk]] & _dummy)`

pytest (https://github.com/pytest-dev/pytest)
- src/_pytest/assertion/rewrite.py:1012:25 error[invalid-assignment] Invalid subscript assignment with key of type `@Todo` and value of type `expr` on object of type `dict[str, str]`
+ src/_pytest/assertion/rewrite.py:1012:25 error[invalid-assignment] Invalid subscript assignment with key of type `Unknown` and value of type `expr` on object of type `dict[str, str]`
- src/_pytest/assertion/rewrite.py:1110:17 error[invalid-assignment] Invalid subscript assignment with key of type `@Todo` and value of type `NamedExpr` on object of type `dict[str, str]`
+ src/_pytest/assertion/rewrite.py:1110:17 error[invalid-assignment] Invalid subscript assignment with key of type `Unknown` and value of type `NamedExpr` on object of type `dict[str, str]`
- src/_pytest/assertion/rewrite.py:1128:21 error[invalid-assignment] Invalid subscript assignment with key of type `@Todo` and value of type `NamedExpr` on object of type `dict[str, str]`
+ src/_pytest/assertion/rewrite.py:1128:21 error[invalid-assignment] Invalid subscript assignment with key of type `Unknown` and value of type `NamedExpr` on object of type `dict[str, str]`

schema_salad (https://github.com/common-workflow-language/schema_salad)
+ src/schema_salad/dotnet_codegen.py:509:45 error[invalid-argument-type] Argument to function `DotNetCodeGen.safe_name` is incorrect: Expected `str`, found `(dict[str, Any] & ~Top[MutableSequence[Unknown]]) | (str & ~Top[MutableSequence[Unknown]])`
+ src/schema_salad/java_codegen.py:571:45 error[invalid-argument-type] Argument to function `JavaCodeGen.safe_name` is incorrect: Expected `str`, found `(dict[str, Any] & ~Top[MutableSequence[Unknown]]) | (str & ~Top[MutableSequence[Unknown]])`
+ src/schema_salad/python_codegen.py:490:40 error[invalid-argument-type] Argument to function `PythonCodeGen.safe_name` is incorrect: Expected `str`, found `(dict[str, Any] & ~Top[MutableSequence[Unknown]]) | (str & ~Top[MutableSequence[Unknown]])`
+ src/schema_salad/typescript_codegen.py:444:40 error[invalid-argument-type] Argument to function `TypeScriptCodeGen.safe_name` is incorrect: Expected `str`, found `(dict[str, Any] & ~Top[MutableSequence[Unknown]]) | (str & ~Top[MutableSequence[Unknown]])`

Full report with detailed diff (timing results)

charliermarsh · 2026-06-22T13:58:31Z

This PR covers sequences; #25941 extends the same behavior to class attributes and mappings.

dhruvmanila

Overall, this change looks good to me. It’s a bit late here, and I’d like to spend more time on a few implementation details, but I don’t think that should block merging. Most of my comments are about expanding the test suite, they may uncover edge cases, but I don’t currently have any concerns to avoid merging this PR.

dhruvmanila · 2026-06-23T12:54:32Z

+def test_incompatible_declared_capture(subject: int) -> None:
+    item: str
+    match subject:
+        case item:  # error: [invalid-assignment]


Unrelated, I like the diagnostic message of mypy in this specific case considering the surrounding context of a pattern capture. This is just something I wanted to highlight and doesn't need to be implemented.

error: Incompatible types in capture pattern (pattern captures type "int", variable has type "str")

dhruvmanila · 2026-06-23T15:41:18Z

Is there a reason that most of these test cases have single case arms? Could we have either an additional case arm, which could just be a catch-all, testing (reveal_type) the other side of the narrowing? Or, do you think it's not worth it?

dhruvmanila · 2026-06-23T15:48:01Z

+def test_constructed_mutable_sequence_alias_does_not_keep_length(
+    value: list[int],
+) -> int:  # error: [invalid-return-type]
+    match value.copy():
+        case [_] as whole:
+            whole.append(2)
+            match whole:
+                case [_]:
+                    return 1
+        case _:
+            raise ValueError


I'm not sure what this test case is testing? The invalid-return-type is coming from the inner match not matching the case expression. Removing the whole.append leads to no change.

Same question for the next test case which is similar but uses nested sequence pattern.

dhruvmanila · 2026-06-23T17:14:51Z

Actually, Codex found a couple of issues:

I think (1) is interesting, (2) will require __match_args__ which I don't think is supported currently (astral-sh/ty#887) and (3) is related to one of the comment that I raised.

[P1] Later cases can narrow a capture instead of the saved match subject.
In builder.rs, original subject bindings are preserved only when the current case rebinds the subject. An earlier guarded case can rebind it and then fail its guard:
```
x = (1,)
match x:
    case [x] if flag:
        pass
    case [1]:
        reveal_type(x)  # ty: Never; runtime: Literal[1]
```
The second predicate is attached to the capture from case one rather than the tuple binding used when the subject was evaluated. This also suppresses an invalid x + "bad" operation. The original subject binding IDs need to be retained for every case.
[P1] Attribute-bearing class patterns are treated as definite matches.
match_pattern.rs subtracts every instance matched by Class(_, Irrefutable). However, an irrefutable child capture does not make attribute extraction irrefutable:
```
class C:
    __match_args__ = ("missing",)

match C():
    case C(item):
        pass
    case remainder:
        reveal_type(remainder)  # ty: Never, but this branch runs
```
Until attribute availability can be proven, argument-bearing class patterns cannot contribute the whole class to definite-match subtraction.
[P1] Mutable aliases retain stale negative sequence-shape facts.
narrow.rs removes positive length/index facts for mutable aliases, but subject_ty can already contain negative shape constraints from earlier cases:
```
match value:
    case []:
        pass
    case whole:
        whole.clear()
        match whole:
            case []:
                reveal_type(whole)  # ty: Never; runtime branch is reached
```
Both positive and negative structural facts must be excluded from the durable type of a mutable alias.

charliermarsh · 2026-06-23T17:16:20Z

(2) is solved by a subsequent PR, intentionally not included here.

charliermarsh · 2026-06-23T17:18:07Z

(I will review the others; I thought I already fixed (1). Thanks!)

charliermarsh added the ty Multi-file analysis & type inference label Jun 13, 2026

charliermarsh mentioned this pull request Jun 13, 2026

[ty] Infer types for names bound in match patterns #25637

Merged

charliermarsh force-pushed the charlie/revive-equality-narrowing branch from a923ba3 to d3b7208 Compare June 13, 2026 12:52

charliermarsh force-pushed the charlie/pattern-as branch from 5721b8d to 7bd8d49 Compare June 13, 2026 12:52

charliermarsh force-pushed the charlie/revive-equality-narrowing branch from d3b7208 to 996f4e8 Compare June 13, 2026 13:33

charliermarsh force-pushed the charlie/pattern-as branch 3 times, most recently from cd30fbb to 5919bfb Compare June 13, 2026 15:40

charliermarsh changed the base branch from charlie/revive-equality-narrowing to charlie/reuse-equality-compatibility June 13, 2026 15:40

charliermarsh mentioned this pull request Jun 13, 2026

[ty] Infer class and mapping pattern bindings #25941

Merged

charliermarsh force-pushed the charlie/pattern-as branch 2 times, most recently from c35e207 to 0636474 Compare June 13, 2026 19:25

charliermarsh force-pushed the charlie/reuse-equality-compatibility branch from 5ac9446 to e9ca170 Compare June 13, 2026 20:10

charliermarsh force-pushed the charlie/pattern-as branch from 0636474 to 4011765 Compare June 13, 2026 20:11

charliermarsh force-pushed the charlie/reuse-equality-compatibility branch 2 times, most recently from 6621128 to 3c57454 Compare June 13, 2026 21:31

charliermarsh force-pushed the charlie/pattern-as branch 2 times, most recently from 2a91303 to 279abdd Compare June 13, 2026 22:33

charliermarsh changed the base branch from charlie/reuse-equality-compatibility to charlie/strict-membership-narrowing June 13, 2026 22:34

charliermarsh force-pushed the charlie/strict-membership-narrowing branch from 2d8ff2b to bc33265 Compare June 14, 2026 02:17

charliermarsh force-pushed the charlie/pattern-as branch 2 times, most recently from f81d5bb to a1661c5 Compare June 14, 2026 03:10

charliermarsh mentioned this pull request Jun 14, 2026

[ty] Narrow match subjects through structural patterns #25942

Open

charliermarsh force-pushed the charlie/pattern-as branch 2 times, most recently from fb131bf to 7340670 Compare June 14, 2026 03:53

charliermarsh force-pushed the charlie/strict-membership-narrowing branch from e3b322e to 02a1ad2 Compare June 14, 2026 04:17

charliermarsh force-pushed the charlie/pattern-as branch from 7340670 to 7a36619 Compare June 14, 2026 04:17

charliermarsh force-pushed the charlie/pattern-as branch from 2292238 to 4f68cc8 Compare June 20, 2026 19:18

charliermarsh changed the base branch from charlie/reuse-equality-compatibility to main June 20, 2026 19:18

charliermarsh force-pushed the charlie/pattern-as branch from 4f68cc8 to a334023 Compare June 22, 2026 13:37

dhruvmanila approved these changes Jun 23, 2026

View reviewed changes

charliermarsh added 22 commits June 23, 2026 13:19

[ty] Infer types for names bound in match patterns

3279156

[ty] Preserve binding identity across OR patterns

ec40563

[ty] Make binding helpers associated functions

85194f5

[ty] Avoid retaining mutable sequence match facts

5d55c13

[ty] Skip bindings for impossible aliases

c0bd076

[ty] Keep invalid class-pattern test focused

9f7265a

[ty] Separate constrained pattern TypeVar analysis

c4046dc

[ty] Clarify successful pattern bindings

36dacd7

[ty] Format pattern binding mdtest

0e778a4

[ty] Keep unsupported OR bindings conservative

f928ae0

[ty] Cover pattern capture declaration scopes

51dfb20

[ty] Keep class pattern alias coverage focused

46abe33

[ty] Avoid unnecessary match fallthrough work

80358bb

[ty] Avoid compounding pattern fallthrough constraints

8116418

[ty] Cover nested captures from TypeVar bounds

6ea252b

[ty] Document match pattern binding helpers

44ad3bd

[ty] Clarify durable sequence binding types

30c15fe

[ty] Speed up enum match fallthrough

8c87411

[ty] Update match binding inlay hint snapshots

86ebd29

[ty] Clean up tests

7bf1924

[ty] Fix match subject narrowing

9cf3e88

[ty] Simplify sequence pattern narrowing

4479750

Uh oh!

Conversation

charliermarsh commented Jun 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Uh oh!

astral-sh-bot Bot commented Jun 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Typing conformance results

Summary

True positives changed (2)

Uh oh!

astral-sh-bot Bot commented Jun 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Memory usage report

Uh oh!

astral-sh-bot Bot commented Jun 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ecosystem-analyzer results

Uh oh!

charliermarsh commented Jun 22, 2026

Uh oh!

dhruvmanila left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dhruvmanila Jun 23, 2026

Choose a reason for hiding this comment

Uh oh!

dhruvmanila Jun 23, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

dhruvmanila Jun 23, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

dhruvmanila commented Jun 23, 2026

Uh oh!

charliermarsh commented Jun 23, 2026

Uh oh!

charliermarsh commented Jun 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

charliermarsh commented Jun 13, 2026 •

edited

Loading

astral-sh-bot Bot commented Jun 13, 2026 •

edited

Loading

astral-sh-bot Bot commented Jun 13, 2026 •

edited

Loading

astral-sh-bot Bot commented Jun 13, 2026 •

edited

Loading

`ecosystem-analyzer` results