[ty] Evaluate reachability of non-definitely-bound to Ambiguous by Glyphack · Pull Request #19579 · astral-sh/ruff

Glyphack · 2025-07-27T21:06:19Z

Summary

If the expression (or any child expressions) is not definitely bound the reachability constraint evaluation is determined as ambiguous.

This fixes the infinite cycles panic in the following code:

from typing import Literal

class Toggle:
    def __init__(self: "Toggle"):
        if not self.x:
            self.x: Literal[True] = True

Credit of this solution is for David.

Test Plan

Added a test case with too many cycle iterations panic.
Previous tests.

github-actions · 2025-07-27T21:10:10Z

`mypy_primer` results

Changes were detected when running on open source projects

trio (https://github.com/python-trio/trio)
+ src/trio/_tests/test_util.py:255:12: error[unresolved-attribute] Type `ModuleType` has no attribute `some_func`
+ src/trio/_tests/test_util.py:256:12: error[unresolved-attribute] Type `ModuleType` has no attribute `some_func`
+ src/trio/_tests/test_util.py:258:12: error[unresolved-attribute] Type `ModuleType` has no attribute `not_funclike`
+ src/trio/_tests/test_util.py:259:12: error[unresolved-attribute] Type `ModuleType` has no attribute `_private`
+ src/trio/_tests/test_util.py:260:12: error[unresolved-attribute] Type `ModuleType` has no attribute `_private`
+ src/trio/_tests/test_util.py:261:12: error[unresolved-attribute] Type `ModuleType` has no attribute `_private`
+ src/trio/_tests/test_util.py:263:12: error[unresolved-attribute] Type `ModuleType` has no attribute `only_has_name`
+ src/trio/_tests/test_util.py:264:12: error[unresolved-attribute] Type `ModuleType` has no attribute `only_has_name`
+ src/trio/_tests/test_util.py:265:24: error[unresolved-attribute] Type `ModuleType` has no attribute `only_has_name`
+ src/trio/_tests/test_util.py:267:12: error[unresolved-attribute] Type `ModuleType` has no attribute `SomeClass`
+ src/trio/_tests/test_util.py:267:12: error[unresolved-attribute] Type `type` has no attribute `method`
+ src/trio/_tests/test_util.py:268:12: error[unresolved-attribute] Type `ModuleType` has no attribute `SomeClass`
+ src/trio/_tests/test_util.py:268:12: error[unresolved-attribute] Type `type` has no attribute `method`
+ src/trio/_tests/test_util.py:269:12: error[unresolved-attribute] Type `ModuleType` has no attribute `SomeClass`
+ src/trio/_tests/test_util.py:269:12: error[unresolved-attribute] Type `type` has no attribute `method`
+ src/trio/_tests/test_util.py:271:5: error[unresolved-attribute] Type `ModuleType` has no attribute `some_func`
+ src/trio/_tests/test_util.py:272:5: error[unresolved-attribute] Type `ModuleType` has no attribute `some_func`
+ src/trio/_tests/test_util.py:273:5: error[unresolved-attribute] Type `ModuleType` has no attribute `_private`
+ src/trio/_tests/test_util.py:274:5: error[unresolved-attribute] Type `ModuleType` has no attribute `SomeClass`
- Found 656 diagnostics
+ Found 675 diagnostics

static-frame (https://github.com/static-frame/static-frame)
+ static_frame/core/util.py:1386:32: warning[possibly-unresolved-reference] Name `rows` used when possibly not defined
- Found 1788 diagnostics
+ Found 1789 diagnostics

dd-trace-py (https://github.com/DataDog/dd-trace-py)
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:65:12: warning[possibly-unbound-attribute] Attribute `vulnerability_budget` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:66:12: warning[possibly-unbound-attribute] Attribute `vulnerabilities_request_limit` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:80:12: warning[possibly-unbound-attribute] Attribute `is_first_vulnerability` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:81:12: warning[possibly-unbound-attribute] Attribute `vulnerability_budget` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:82:12: warning[possibly-unbound-attribute] Attribute `vulnerabilities_request_limit` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:83:12: warning[possibly-unbound-attribute] Attribute `vulnerabilities_request_limit` on type `IASTEnvironment | None` is possibly unbound
+ tests/appsec/iast/taint_sinks/test_vulnerability_detection.py:84:12: warning[possibly-unbound-attribute] Attribute `vulnerabilities_request_limit` on type `IASTEnvironment | None` is possibly unbound
- Found 6595 diagnostics
+ Found 6602 diagnostics

sympy (https://github.com/sympy/sympy)
+ sympy/polys/numberfields/galoisgroups.py:126:21: warning[possibly-unresolved-reference] Name `current_degree` used when possibly not defined
- Found 13401 diagnostics
+ Found 13402 diagnostics

scipy (https://github.com/scipy/scipy)
+ scipy/optimize/tests/test_bracket.py:318:9: error[unresolved-attribute] Unresolved attribute `count` on type `def f(x) -> Unknown`.
- Found 6420 diagnostics
+ Found 6421 diagnostics

No memory usage changes detected ✅

codspeed-hq · 2025-07-27T21:19:11Z

CodSpeed WallTime Performance Report

Merging #19579 will not alter performance

_{Comparing Glyphack:reachability-ambiguous-not-definitely-bound (c4ce73a) with main (18eaa65)}

Summary

✅ 8 untouched benchmarks

codspeed-hq · 2025-07-27T21:19:32Z

CodSpeed Instrumentation Performance Report

Merging #19579 will not alter performance

_{Comparing Glyphack:reachability-ambiguous-not-definitely-bound (c4ce73a) with main (18eaa65)}

Summary

✅ 42 untouched benchmarks

github-actions · 2025-07-29T15:19:45Z

Diagnostic diff on typing conformance tests

No changes detected when running ty on typing conformance tests ✅

crates/ty_python_semantic/src/types/class.rs

…s-not-definitely-bound

sharkdp · 2025-08-27T13:11:11Z

I'm looking into making the changes described above.

crates/ty_python_semantic/src/types/infer.rs

github-actions · 2025-08-27T15:45:10Z

`ecosystem-analyzer` results

Lint rule	Added	Removed	Changed
`unresolved-attribute`	20	0	0
`possibly-unbound-attribute`	7	0	0
`possibly-unresolved-reference`	2	0	0
Total	29	0	0

…tions (#20114) ## Summary Properly preserve type qualifiers when accessing attributes on unions and intersections. This is a prerequisite for #19579. Also fix a completely wrong implementation of `map_with_boundness_and_qualifiers`. It now closely follows `map_with_boundness` (just above). ## Test Plan I thought about it, but didn't find any easy way to test this. This only affected `Type::member`. Things like validation of attribute writes (where type qualifiers like `ClassVar` and `Final` are important) were already handling things correctly.

…s-not-definitely-bound

sharkdp · 2025-08-28T11:59:39Z

I tried various other solutions here, and non of them worked as well as what @Glyphack implemented in the first place. The main problem is that we can only build unions/intersections of types. We can't build unions/intersections of TypeQualifiers and boundness information, which requires us to merge them somehow when we operate on unions/intersections (for example, when we access an attribute on a union/intersection). With this representation, it is impossible to distinguish between a case where we have C | None (where an attribute access should definitely yield possibly-unbound, due to the None in the union), and something like C | Unknown (where we would like to silence the possibly-unbound-attribute diagnostic if the attribute is only possibly bound inside a method of C).

I also tried various completely different approaches of solving this problem, and none of them hit the right granularity. Either they are not able to solve the Toggle problem, or they are too big of a hammer, and lead to insufficient reachability analysis in real world use cases.

So I am returning this PR to a previous state (with a couple of smaller cleanups, renamings, and additional tests). I am still not very satisfied with the fact that we track boundness information in a TypeQualifier, but I'm out of ideas. And we would very much like to proceed with type-of-self.

sharkdp

Thank you again, @Glyphack!

AlexWaygood

Thank you!

crates/ty_python_semantic/resources/mdtest/statically_known_branches.md

crates/ty_python_semantic/src/place.rs

crates/ty_python_semantic/src/types.rs

crates/ty_python_semantic/src/types/infer.rs

* main: Fix mdtest ignore python code blocks (#20139) [ty] add support for cyclic legacy generic protocols (#20125) [ty] add cycle detection for find_legacy_typevars (#20124) Use new diff rendering format in tests (#20101) [ty] Fix 'too many cycle iterations' for unions of literals (#20137) [ty] No boundness analysis for implicit instance attributes (#20128) Bump 0.12.11 (#20136) [ty] Benchmarks for problematic implicit instance attributes cases (#20133) [`pyflakes`] Fix `allowed-unused-imports` matching for top-level modules (`F401`) (#20115) Move GitLab output rendering to `ruff_db` (#20117) [ty] Evaluate reachability of non-definitely-bound to Ambiguous (#19579) [ty] Introduce a representation for the top/bottom materialization of an invariant generic (#20076) [`flake8-async`] Implement `blocking-http-call-httpx` (`ASYNC212`) (#20091) [ty] print diagnostics with fully qualified name to disambiguate some cases (#19850) [`ruff`] Preserve relative whitespace in multi-line expressions (`RUF033`) (#19647)

…tions (astral-sh#20114) ## Summary Properly preserve type qualifiers when accessing attributes on unions and intersections. This is a prerequisite for astral-sh#19579. Also fix a completely wrong implementation of `map_with_boundness_and_qualifiers`. It now closely follows `map_with_boundness` (just above). ## Test Plan I thought about it, but didn't find any easy way to test this. This only affected `Type::member`. Things like validation of attribute writes (where type qualifiers like `ClassVar` and `Final` are important) were already handling things correctly.

…al-sh#19579) ## Summary closes astral-sh/ty#692 If the expression (or any child expressions) is not definitely bound the reachability constraint evaluation is determined as ambiguous. This fixes the infinite cycles panic in the following code: ```py from typing import Literal class Toggle: def __init__(self: "Toggle"): if not self.x: self.x: Literal[True] = True ``` Credit of this solution is for David. ## Test Plan - Added a test case with too many cycle iterations panic. - Previous tests. --------- Co-authored-by: David Peter <[email protected]>

…nalysis (#22971) ## Summary cf: #19579, #20566 (comment) Currently, we use the query `static_expression_truthiness` to determine the truthiness of an expression. This always determines the truthinesses of non-definitely-bound places as `Ambiguous`, preventing cycles that occur when their reachabilities depend on themselves. However, this can lead to inaccurate analysis of code like the following. ```python def _(flag: bool): if flag: ALWAYS_TRUE_IF_BOUND = True # error: [possibly-unresolved-reference] "Name `ALWAYS_TRUE_IF_BOUND` used when possibly not defined" if ALWAYS_TRUE_IF_BOUND: x = 1 else: x = 2 # If `ALWAYS_TRUE_IF_BOUND` were not defined, an error would occur, and therefore the `x = 2` branch would never be executed. reveal_type(x) # expected: Literal[1], but: Literal[1, 2] ``` Since #20566, we can determine the `Place` by looking at the previous cycle result. Therefore, we can now remove `static_expression_truthiness`, improving reachability analysis of the above code. ## Test Plan mdtest updated

AlexWaygood added the ty Multi-file analysis & type inference label Jul 27, 2025

AlexWaygood changed the title ~~Evaluate reachability of non-definitely-bound to Ambiguous~~ [ty] Evaluate reachability of non-definitely-bound to Ambiguous Jul 27, 2025

This comment was marked as resolved.

Sign in to view

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch from d5dafc5 to 3430eea Compare July 29, 2025 15:17

This comment was marked as resolved.

Sign in to view

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch 4 times, most recently from d383152 to b05868f Compare August 4, 2025 18:37

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch from b05868f to 9be1e82 Compare August 16, 2025 11:47

Glyphack added 5 commits August 18, 2025 20:56

Evaluate reachability of non-definitely-bound to Ambiguos

8ae9984

Check definitely bound using a visitor

c8f5658

Use record_place_load

2403cca

Remove visitor version

165a68c

Record unbound in attrs after fallback mechanism

93ea0bb

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch from 9be1e82 to feca174 Compare August 18, 2025 22:31

This comment was marked as resolved.

Sign in to view

Preserve possibly unbound place during instance member lookup

590ea0d

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch from feca174 to 590ea0d Compare August 19, 2025 17:38

Remove record_place_load to consider fallbacks

75a3dc7

Glyphack force-pushed the reachability-ambiguous-not-definitely-bound branch from aa650e9 to a84b8ce Compare August 20, 2025 13:43

sharkdp reviewed Aug 20, 2025

View reviewed changes

crates/ty_python_semantic/src/types/class.rs Show resolved Hide resolved

Glyphack added 4 commits August 20, 2025 20:03

Determine boundness in infer_attribute_load for implicit attributes

36e8c82

Cargo clippy

a84b8ce

Move tests

0ae4493

Handle non-declared attributes

f8626ce

Merge remote-tracking branch 'origin/main' into reachability-ambiguou…

66f3563

…s-not-definitely-bound

sharkdp reviewed Aug 27, 2025

View reviewed changes

crates/ty_python_semantic/src/types/infer.rs Outdated Show resolved Hide resolved

sharkdp force-pushed the reachability-ambiguous-not-definitely-bound branch from 4c1da8c to 2218db4 Compare August 27, 2025 14:31

sharkdp mentioned this pull request Aug 27, 2025

[ty] Preserve qualifiers when accessing attributes on unions/intersections #20114

Merged

sharkdp force-pushed the reachability-ambiguous-not-definitely-bound branch from 94d1897 to 71bd443 Compare August 27, 2025 15:15

sharkdp added the ecosystem-analyzer label Aug 27, 2025

sharkdp added ecosystem-analyzer and removed ecosystem-analyzer labels Aug 28, 2025

sharkdp added 2 commits August 28, 2025 13:51

Cleanup, tests

ee1cb2c

Merge remote-tracking branch 'origin/main' into reachability-ambiguou…

a18a7ce

…s-not-definitely-bound

sharkdp force-pushed the reachability-ambiguous-not-definitely-bound branch from 966553c to a18a7ce Compare August 28, 2025 11:51

sharkdp added ecosystem-analyzer and removed ecosystem-analyzer labels Aug 28, 2025

Fix comment

3837b2b

sharkdp marked this pull request as ready for review August 28, 2025 12:02

sharkdp approved these changes Aug 28, 2025

View reviewed changes

AlexWaygood approved these changes Aug 28, 2025

View reviewed changes

Review findings

c4ce73a

sharkdp merged commit d9aaacd into astral-sh:main Aug 28, 2025
38 checks passed

Glyphack deleted the reachability-ambiguous-not-definitely-bound branch October 22, 2025 15:47

mtshiba mentioned this pull request Nov 22, 2025

[ty] handle recursive type inference properly #20566

Merged

mtshiba mentioned this pull request Jan 30, 2026

[ty] remove static_expression_truthiness and improve reachability analysis #22971

Merged

Comments

Conversation

Glyphack commented Jul 27, 2025 • edited by sharkdp Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test Plan

Uh oh!

github-actions bot commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

mypy_primer results

Uh oh!

codspeed-hq bot commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CodSpeed WallTime Performance Report

Merging #19579 will not alter performance

Summary

Uh oh!

codspeed-hq bot commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CodSpeed Instrumentation Performance Report

Merging #19579 will not alter performance

Summary

Uh oh!

This comment was marked as resolved.

This comment was marked as resolved.

github-actions bot commented Jul 29, 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!

This comment was marked as resolved.

Uh oh!

sharkdp commented Aug 27, 2025

Uh oh!

Uh oh!

github-actions bot commented Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ecosystem-analyzer results

Uh oh!

sharkdp commented Aug 28, 2025

Uh oh!

sharkdp left a comment

Choose a reason for hiding this comment

Uh oh!

AlexWaygood left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Glyphack commented Jul 27, 2025 •

edited by sharkdp

Loading

github-actions bot commented Jul 27, 2025 •

edited

Loading

`mypy_primer` results

codspeed-hq bot commented Jul 27, 2025 •

edited

Loading

codspeed-hq bot commented Jul 27, 2025 •

edited

Loading

github-actions bot commented Jul 29, 2025 •

edited

Loading

github-actions bot commented Aug 27, 2025 •

edited

Loading

`ecosystem-analyzer` results