Introduce `ClosureType::isStaticClosure()` by phpstan-bot · Pull Request #5699 · phpstan/phpstan-src

phpstan-bot · 2026-05-18T15:21:56Z

Summary

Introduces isStaticClosure(): TrinaryLogic on ClosureType and CallableParametersAcceptor to track whether a closure is declared with the static keyword. This enables PHPStan to distinguish static from non-static closures in the type system, which is useful for rules that check closure binding compatibility and for narrowing the return types of Closure::bindTo() and Closure::bind().

Changes

Core type system

Added isStatic parameter (defaulting to Maybe) to ClosureType constructor (src/Type/ClosureType.php)
Added isStaticClosure(): TrinaryLogic method to ClosureType
Updated ClosureType::equals() to compare isStatic
Updated ClosureType::describe() to show static prefix at precise level and static/non-static at cache level
Updated ClosureType::traverse() and traverseSimultaneously() to preserve isStatic

Interface and implementations

Added isStaticClosure(): TrinaryLogic to CallableParametersAcceptor interface (src/Reflection/Callables/CallableParametersAcceptor.php)
Implemented in all concrete classes:
- CallableType — returns Maybe (callable could be anything)
- TrivialParametersAcceptor — returns Maybe
- FunctionCallableVariant — returns No (named functions are not closures)
- InaccessibleMethod — returns No
- ExtendedCallableFunctionVariant — propagated via constructor parameter
- ResolvedFunctionVariantWithCallable — propagated via constructor parameter

Closure creation

Updated ClosureTypeResolver to pass TrinaryLogic::createFromBoolean($expr->static) at all three ClosureType construction sites (src/Analyser/ExprHandler/Helper/ClosureTypeResolver.php)
Updated InitializerExprTypeResolver to set isStatic: TrinaryLogic::createYes() for static closures in constant contexts (src/Reflection/InitializerExprTypeResolver.php)

Propagation

ParametersAcceptorSelector — propagates isStaticClosure when combining/wrapping callable variants (src/Reflection/ParametersAcceptorSelector.php)
GenericParametersAcceptorResolver — propagates through template resolution (src/Reflection/GenericParametersAcceptorResolver.php)
ClosureFromCallableDynamicReturnTypeExtension — propagates from variant to new ClosureType
TypeNodeResolver — propagates when overriding ClosureType with PHPDoc
RuleLevelHelper — propagates when transforming closure types
CallableTypeHelper::isParametersAcceptorSuperTypeOf() — checks isStaticClosure compatibility

Root cause

The issue requested tracking whether a closure is static in the type system. Previously, the static keyword on closures was only tracked at the AST level ($expr->static) and used for scope resolution (preventing $this access), but was not represented in ClosureType. This meant rules and return type extensions that needed to know if a closure was static had to inspect the AST node directly rather than querying the type.

The fix adds isStatic: TrinaryLogic to ClosureType and propagates it through all relevant interfaces and creation sites. The TrinaryLogic three-valued logic (yes/no/maybe) is used because closures from PHPDoc annotations or generic type resolution may have unknown static-ness.

Test

Type inference test (tests/PHPStan/Analyser/nsrt/closure-static-type.php): Verifies static prefix in type descriptions for static closures and arrow functions, verifies Closure::bindTo() and Closure::bind() return null for static closures and Closure for non-static closures, and verifies Closure|null for closures with unknown static-ness.
Unit tests (tests/PHPStan/Type/ClosureTypeTest.php): Tests for isSuperTypeOf() with static/non-static/maybe-static closure combinations, equals() comparisons, describe() output at all verbosity levels, and isStaticClosure() method.
Updated existing tests: degrade-closures.php, bug-7031.php, bug-9764.php, bug-14324.php — updated to expect static prefix for static closures. bug-4734 test updated to expect the new error from calling null returned by Closure::bind() on a static closure.

Refs phpstan/phpstan#14639

VincentLanglet

Remove all the change to ClosureBindDynamicReturnTypeExtension and ClosureBindToDynamicReturnTypeExtension.

This will be done in another PR

VincentLanglet · 2026-05-19T21:29:17Z

+				return $prefix . $printer->print($selfWithoutParameterNames->toPhpDocNode());
+			},
+			function (): string {
+				$prefix = !$this->isStatic->maybe() ? ($this->isStatic->yes() ? 'static ' : 'non-static ') : '';


There is no reason to use non-static here, we don't use non-pure.

Both commits are pushed. Here's a summary of the changes:

Commit 1: Revert Closure::bindTo()/Closure::bind() return type narrowing

Reverted ClosureBindDynamicReturnTypeExtension and ClosureBindToDynamicReturnTypeExtension to their original state (no static closure checks)

Updated closure-static-type.php test assertions to expect the closure type (not null) from bind/bindTo

Reverted bug-4734 integration test to original 5-error expectation

Commit 2: Use hyphen prefix and extract helper method

Changed prefix from 'static ' to 'static-' (matching pure-Closure convention)

Removed non-static prefix at cache level (consistent with not using non-pure)

Extracted describeWithPrefix() private method, eliminating the duplicated closure body across value, precise, and cache callbacks (cache now falls through to precise)

Updated all test assertion strings from 'static Closure' to 'static-Closure'

VincentLanglet · 2026-05-19T21:31:07Z


 				return $printer->print($selfWithoutParameterNames->toPhpDocNode());
 			},
+			function (): string {


Maybe a private methode could be created to refactor the value and precise level

VincentLanglet · 2026-05-19T21:31:47Z

 				return $printer->print($selfWithoutParameterNames->toPhpDocNode());
 			},
+			function (): string {
+				$prefix = $this->isStatic->yes() ? 'static ' : '';


prefix should be with a -

VincentLanglet · 2026-05-19T22:16:02Z

+			fn (): string => $this->describeBody(showPure: true, showStatic: false),
+			fn (): string => $this->describeBody(showPure: true, showStatic: true),


Do not use namedParam

Done. Removed the named parameters from the describeBody() calls as requested by the reviewer. Tests and static analysis both pass.

VincentLanglet

Some AttributeReflectionTest::testAttributeReflections test are failing in PHP 8.5 because the expected output is static-Closure rather than Closure

phpstan-bot · 2026-05-19T22:32:08Z

Pushed. Updated the 6 test expectations in AttributeReflectionTest where static closures in PHP 8.5 attributes now describe as static-Closure(...) instead of Closure(...) at precise verbosity level. All tests and static analysis pass.

staabm · 2026-05-25T07:00:56Z

@VincentLanglet this PR does not implement the use-cases for isStatic() described in phpstan/phpstan#14639 (therefore cannot close the issue). do you intend to implement the use-cases in a followup?

VincentLanglet · 2026-05-25T07:06:22Z

@VincentLanglet this PR does not implement the use-cases for isStatic() described in phpstan/phpstan#14639 (therefore cannot close the issue). do you intend to implement the use-cases in a followup?

Yes, I prefer to split the PR because it already big enough Imho

ondrejmirtes

I also miss some "accepts" tests.

And I feel like that will show that we need to update getRecommendedLevelByType. When a non-static closure is rejected by static closure, the verbosity level should be increased to precise (very-verbose).

…o()`/`Closure::bind()` return type for static closures - Add `isStaticClosure(): TrinaryLogic` to `CallableParametersAcceptor` interface - Add `isStatic` parameter to `ClosureType` constructor, propagated through `equals()`, `describe()`, `traverse()`, `traverseSimultaneously()` - Show `static` prefix in `describe()` at precise level when closure is definitely static; show `non-static`/`static` at cache level when known - Implement `isStaticClosure()` in all `CallableParametersAcceptor` implementations: `CallableType` (maybe), `TrivialParametersAcceptor` (maybe), `FunctionCallableVariant` (no), `InaccessibleMethod` (no), `ExtendedCallableFunctionVariant` and `ResolvedFunctionVariantWithCallable` (propagated from source) - Set `isStatic` from `$expr->static` in `ClosureTypeResolver` for both `Closure` and `ArrowFunction` AST nodes - Add `isStaticClosure` check to `CallableTypeHelper::isParametersAcceptorSuperTypeOf()` - Propagate `isStaticClosure` through `ParametersAcceptorSelector`, `GenericParametersAcceptorResolver`, `ClosureFromCallableDynamicReturnTypeExtension`, `TypeNodeResolver`, and `RuleLevelHelper` - Narrow `Closure::bindTo()` to return `null` for static closures and `Closure|null` for maybe-static closures - Narrow `Closure::bind()` to return `null` for static closures and `Closure|null` for maybe-static closures - Update existing tests to expect `static` prefix in closure type descriptions

…r static closures This will be done in a separate PR. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…` helper - Change prefix from `static ` to `static-` to match the `pure-Closure` convention - Remove `non-static` prefix at cache level (consistent with not using `non-pure`) - Extract `describeWithPrefix()` private method to deduplicate value/precise callbacks Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…ic` booleans Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…tributes Static closures in PHP 8.5 attributes now correctly describe as `static-Closure(...)` at precise verbosity level. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Consistent with how pure- prefix is handled: the static- prefix only appears for common callable closures (e.g. static-Closure, static-pure-Closure), not for closures with explicit parameters (e.g. Closure(): void stays the same regardless of static-ness). Added tests for the static-pure-Closure combination. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…llable branch Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…isStatic - Add precise callback to ClosureType::describe() that shows static- prefix for all closures (not just common callable) when isStatic is yes - Extract describeCallable() private method to avoid duplication - Add static- prefix to toPhpDocNode() for common callable closures - Add static-Closure/static-pure-Closure parsing in TypeNodeResolver - Pass isStatic to new self() in describe() and to all relevant new ClosureType construction sites (InitializerExprTypeResolver, ClosureTypeFactory) - Update getRecommendedLevelByType to bump to precise level when a ClosureType has definite isStatic (yes or no) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…r static- prefix - Add dataAccepts/testAccepts to ClosureTypeTest with static/non-static/ maybe-static combinations - Add TypeCombinator::union and ::intersect tests for static closures - Add toPhpDocNode tests for static-Closure and static-pure-Closure - Update existing test expectations to include static- prefix at precise level for static closures in closure-static-type, bug-7031, bug-9764, degrade-closures, bug-14324 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

ReflectionFunction::isStatic() does not reliably report static closures on PHP < 8.1, so remove the isStatic parameter from the fromClosureObject() call site. This falls back to TrinaryLogic::createMaybe() (the default), which is safe for this niche runtime-reflection code path. The main analysis path through ClosureTypeResolver correctly determines isStatic from the AST. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

ondrejmirtes · 2026-06-05T07:16:07Z

Still some failures on PHP 8.5.

VincentLanglet · 2026-06-05T08:37:43Z

Still some failures on PHP 8.5.

PHP8.5 tests updated @ondrejmirtes

VincentLanglet requested changes May 19, 2026

View reviewed changes

VincentLanglet reviewed May 19, 2026

View reviewed changes

Comment thread src/Type/ClosureType.php Outdated

VincentLanglet reviewed May 19, 2026

View reviewed changes

VincentLanglet changed the title ~~Introduce ClosureType::isStaticClosure() and narrow Closure::bindTo()/Closure::bind() return type for static closures~~ Introduce ClosureType::isStaticClosure() May 19, 2026

VincentLanglet reviewed May 19, 2026

View reviewed changes

VincentLanglet approved these changes May 19, 2026

View reviewed changes

VincentLanglet requested a review from staabm May 24, 2026 21:08

staabm reviewed May 25, 2026

View reviewed changes

Comment thread src/Type/ClosureType.php Outdated

staabm reviewed May 25, 2026

View reviewed changes

Comment thread src/Type/ClosureType.php Outdated

staabm reviewed May 25, 2026

View reviewed changes

Comment thread src/Type/ClosureType.php

staabm force-pushed the create-pull-request/patch-hguxjyq branch from a05f5a2 to 0786e7a Compare May 25, 2026 08:04

staabm requested a review from VincentLanglet May 25, 2026 08:36

staabm approved these changes May 25, 2026

View reviewed changes

staabm force-pushed the create-pull-request/patch-hguxjyq branch from 0786e7a to 9486a3f Compare May 27, 2026 08:23

ondrejmirtes requested changes May 27, 2026

View reviewed changes

Comment thread src/Type/ClosureType.php

Comment thread tests/PHPStan/Analyser/nsrt/closure-static-type.php Outdated

Comment thread tests/PHPStan/Type/ClosureTypeTest.php

VincentLanglet reviewed May 27, 2026

View reviewed changes

Comment thread src/Type/ClosureTypeFactory.php Outdated

VincentLanglet changed the base branch from 2.1.x to 2.2.x May 29, 2026 14:03

VincentLanglet force-pushed the create-pull-request/patch-hguxjyq branch from 1dd9c0a to 50a1b44 Compare May 29, 2026 14:14

VincentLanglet self-assigned this May 30, 2026

VincentLanglet requested a review from ondrejmirtes June 3, 2026 21:59

VincentLanglet and others added 7 commits June 5, 2026 09:05

Revert Closure::bindTo()/Closure::bind() return type narrowing fo…

834d066

…r static closures This will be done in a separate PR. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Replace describeWithPrefix() string param with showPure/`showStat…

0d60da5

…ic` booleans Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Remove named parameters from describeBody() calls

b324b57

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Update AttributeReflectionTest expectations for static closures in at…

1acd202

…tributes Static closures in PHP 8.5 attributes now correctly describe as `static-Closure(...)` at precise verbosity level. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Add test cases for static-pure-Closure describe output

67988f6

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

phpstan-bot and others added 6 commits June 5, 2026 09:05

Revert describeBody() extraction, inline static- prefix in isCommonCa…

3a9c3e0

…llable branch Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

simplify

4842943

ondrejmirtes force-pushed the create-pull-request/patch-hguxjyq branch from 50a1b44 to e376c3a Compare June 5, 2026 07:05

Fix

fecd47e

		fn (): string => $this->describeBody(showPure: true, showStatic: false),
		fn (): string => $this->describeBody(showPure: true, showStatic: true),

Conversation

phpstan-bot commented May 18, 2026 • edited by staabm Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Core type system

Interface and implementations

Closure creation

Propagation

Root cause

Test

Uh oh!

VincentLanglet left a comment

Choose a reason for hiding this comment

Uh oh!

VincentLanglet May 19, 2026

Choose a reason for hiding this comment

Uh oh!

phpstan-bot May 19, 2026

Choose a reason for hiding this comment

Uh oh!

VincentLanglet May 19, 2026

Choose a reason for hiding this comment

Uh oh!

VincentLanglet May 19, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

VincentLanglet May 19, 2026

Choose a reason for hiding this comment

Uh oh!

phpstan-bot May 19, 2026

Choose a reason for hiding this comment

Uh oh!

VincentLanglet left a comment

Choose a reason for hiding this comment

Uh oh!

phpstan-bot commented May 19, 2026

Uh oh!

Uh oh!

Uh oh!

staabm commented May 25, 2026

Uh oh!

VincentLanglet commented May 25, 2026

Uh oh!

Uh oh!

ondrejmirtes left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ondrejmirtes commented Jun 5, 2026

Uh oh!

VincentLanglet commented Jun 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

phpstan-bot commented May 18, 2026 •

edited by staabm

Loading