[W.I.P.] Fix implicit to and from bool type juggling #60890

Girgias · 2025-06-24T12:28:30Z

Q	A
Branch?	7.4 (but can rebase to lower)
Bug fix?	maybe
New feature?	no
Deprecations?	no
Issues	N/A
License	MIT

This is to see part of the impact of php/php-src#18879, and it seems to be finding some bugs in tests or SF itself, but I would need confirmation from other people.

This is very much W.I.P. and I might use CI to check that my changes are correct as I have some local test failures even just running with 8.4.8

I am well aware the commit messages are kinda useless at the moment, but they are somewhat self-contained, so they can be looked at individually.

carsonbot · 2025-06-24T12:28:33Z

Hey!

To help keep things organized, we don't allow "Draft" pull requests. Could you please click the "ready for review" button or close this PR and open a new one when you are done?

Note that a pull request does not have to be "perfect" or "ready for merge" when you first open it. We just want it to be ready for a first review.

Cheers!

Carsonbot

…alse

…first arg

stof · 2025-06-24T13:06:31Z

src/Symfony/Component/DependencyInjection/Loader/YamlFileLoader.php

@@ -568,7 +568,7 @@ private function parseDefinition(string $id, array|string|null $service, string
                if (isset($call['method']) && \is_string($call['method'])) {
                    $method = $call['method'];
                    $args = $call['arguments'] ?? [];
-                    $returnsClone = $call['returns_clone'] ?? false;
+                    $returnsClone = $call['returns_clone'] ? true : false;


this will break if returns_clone is not defined in the array.

and that's a case where we expect a boolean if the key is defined, or an undefined key (which should then default to false)

stof · 2025-06-24T13:09:59Z

src/Symfony/Component/DependencyInjection/Definition.php

@@ -307,7 +307,7 @@ public function setMethodCalls(array $calls = []): static
    {
        $this->calls = [];
        foreach ($calls as $call) {
-            $this->addMethodCall($call[0], $call[1], $call[2] ?? false);
+            $this->addMethodCall($call[0], $call[1], $call[2] ? true : false);


this is a case where the third value in the array is expected to be a boolean or absent, which is broken in this change.

stof · 2025-06-24T13:12:21Z

src/Symfony/Component/DependencyInjection/Loader/XmlFileLoader.php

+            $definition->addMethodCall(
+                $call->getAttribute('method'),
+                $this->getArgumentsAsPhp($call, 'argument', $file),
+                (bool) XmlUtils::phpize($call->getAttribute('returns-clone'))


I suggest doing a method handling of the case of the missing attribute (instead of handling it as $call->getAttribute('returns-clone') returning an empty string which then casts to false)

stof · 2025-06-24T13:12:53Z

src/Symfony/Component/DependencyInjection/Loader/YamlFileLoader.php

@@ -589,7 +589,7 @@ private function parseDefinition(string $id, array|string|null $service, string
                    } else {
                        $method = $call[0];
                        $args = $call[1] ?? [];
-                        $returnsClone = $call[2] ?? false;
+                        $returnsClone = $call[2] ? true : false;


the existing code should probably be kept here as well.

stof · 2025-06-24T13:21:57Z

src/Symfony/Component/ErrorHandler/ErrorHandler.php

@@ -402,7 +402,7 @@ public function handleError(int $type, string $message, string $file, int $line)

        // Never throw on warnings triggered by assert()
        if (\E_WARNING === $type && 'a' === $message[0] && 0 === strncmp($message, 'assert(): ', 10)) {
-            $throw = 0;
+            $throw = false;


$throw is still an int if we don't enter this if branch, as it is defined with bitwise comparisons on line 400. It needs to be fixed there as well.

stof · 2025-06-24T13:23:51Z

Some of those changes are definitely legitimate as they are clearly mistakes in the existing code that went unnoticed (static analysis could catch them, but our CI only reports new violations in each PR and we never worked on fixing issues reported in existing code):

the wrong parameter passed in the instantiation of Isbn
the wrong type passed to IntlTestHelper
the wrong type passed to stream_set_blocking
the wrong parameter type in the internal method of GraphvizDumper

Those could definitely be submitted directly (against the 6.4 branch).

We have other cases (when using preg_match or bitwise comparisons) where we indeed rely on the implicit type juggling of the return type, which are legitimate changes once PHP deprecates those type juggling rules (static analysis probably also prevents them for new code).
As we will want to support PHP 8.5 in the 6.4 branch, fixes for them should also be submitted against 6.4 (I would suggest separating them from the ones fixing clear mistakes).

As reported in my review comments, the changes done in the DI component look wrong to me (and are the cause of failures in the CI as well)

Girgias · 2025-06-24T13:28:22Z

Some of those changes are definitely legitimate as they are clearly mistakes in the existing code that went unnoticed (static analysis could catch them, but our CI only reports new violations in each PR and we never worked on fixing issues reported in existing code):
* the wrong parameter passed in the instantiation of `Isbn`

* the wrong type passed to `IntlTestHelper`

* the wrong type passed to `stream_set_blocking`

* the wrong parameter type in the internal method of GraphvizDumper
Those could definitely be submitted directly (against the 6.4 branch).

We have other cases (when using preg_match or bitwise comparisons) where we indeed rely on the implicit type juggling of the return type, which are legitimate changes once PHP deprecates those type juggling rules (static analysis probably also prevents them for new code). As we will want to support PHP 8.5 in the 6.4 branch, fixes for them should also be submitted against 6.4 (I would suggest separating them from the ones fixing clear mistakes).

As reported in my review comments, the changes done in the DI component look wrong to me (and are the cause of failures in the CI as well)

Yes I had figured out the cause, but I was having some other local failures before that made it difficult to waddle through stuff. There is also a bug in master on php-src which is not helping me to get a good overview of the situation.

Question about Isbn should I drop the null argument as it's the default or not?

stof · 2025-06-24T13:37:14Z

src/Symfony/Component/DependencyInjection/Tests/Compiler/CheckTypeDeclarationsPassTest.php

@@ -336,7 +336,7 @@ public function testProcessFailsOnPassingClassToScalarTypedParameter()
        (new CheckTypeDeclarationsPass(true))->process($container);
    }

-    public function testProcessSuccessOnPassingBadScalarType()
+    public function xtestProcessSuccessOnPassingBadScalarType()


why skipping this test ?

I stole this from @nicolas-grekas WIP commit: nicolas-grekas@02e3881

stof · 2025-06-24T13:40:28Z

src/Symfony/Bundle/TwigBundle/Tests/Functional/AttributeExtensionTest.php

@@ -51,7 +51,7 @@ public function registerContainerConfiguration(LoaderInterface $loader): void
                    $container->register(StaticExtensionWithAttributes::class, StaticExtensionWithAttributes::class)
                        ->setAutoconfigured(true);
                    $container->register(RuntimeExtensionWithAttributes::class, RuntimeExtensionWithAttributes::class)
-                        ->setArguments(['prefix_'])
+                        ->setArguments([true])


given that RuntimeExtensionWithAttributes performs string concatenation with the prefix, the right fix is rather to change the constructor argument type in this RuntimeExtensionWithAttributes stub class.

stof · 2025-06-24T14:18:33Z

src/Symfony/Component/String/LazyString.php

@@ -76,7 +76,8 @@ final public static function isStringable(mixed $value): bool
     */
    final public static function resolve(\Stringable|string|int|float|bool $value): string
    {
-        return $value;
+        // TODO Do something more useful for false and true?


the TODO should be removed, as a string cast is also what we do in fromStringable

…tead of 0

Not sure what those asserts are doing however

There is no need to test false as this is a PHP engine behaviour to cast to empty string

Girgias added 6 commits June 24, 2025 13:39

Pull fixes from Nicolas

1ac3888

Pull ignores/suppressed tests from Nicolas

2db7dac

Use false instead of 0

9e73d6a

IntlTestHelper::requireFullIntl() use null as second arg instead of f…

d388344

…alse

Check return value of getenv()

4923fa4

Symfony\Component\Validator\Constraints\Isbn does not take a bool as …

1cefb6b

…first arg

stof reviewed Jun 24, 2025

View reviewed changes

carsonbot closed this Jun 24, 2025

nicolas-grekas reopened this Jun 24, 2025

Girgias force-pushed the bool-type-juggling branch from d89795e to b812e26 Compare June 24, 2025 13:33

Girgias marked this pull request as ready for review June 24, 2025 13:35

Girgias requested review from lyrixx, yceruto, kbond, dunglas and xabbuh as code owners June 24, 2025 13:35

carsonbot added the Status: Needs Review label Jun 24, 2025

carsonbot added this to the 7.4 milestone Jun 24, 2025

symfony deleted a comment from carsonbot Jun 24, 2025

stof reviewed Jun 24, 2025

View reviewed changes

Girgias added 4 commits June 25, 2025 14:23

intl false instead of null

85b52c5

Cast result of preg_match to bool

1f4dd46

Use bool as param type

d054d65

Cast bitwise op result to bool

65c324a

Girgias added 12 commits June 25, 2025 14:23

ErrorHandler::handleError() cast bitwise op to bool and use false ins…

b5a80da

…tead of 0

Yield empty string when no doc comment exists

1b7ac21

Set value to null if parse_url() fails badly and returns false

e2a60d3

Fix deprecation with PHPUnit's assertTrue

d13a389

Not sure what those asserts are doing however

Remove ZPP test

49f8c6d

There is no need to test false as this is a PHP engine behaviour to cast to empty string

file_get_contents() does not take a fopen mode parameter

a007578

Symfony\Component\String\LazyString::resolve() add explicit string cast

048257f

Fix string to true type juggling in test

105d6f1

Remove ZPP test

177bdb8

There is no need to test false as this is a PHP engine behaviour to cast to empty string

Use empty string instead of false

94ba9ae

Explicit cast to bool from string

93d90b6

Definition::addMethodCall() arg 3 must be a bool v2

a249be4

Girgias force-pushed the bool-type-juggling branch from b812e26 to a249be4 Compare June 25, 2025 13:23

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

[W.I.P.] Fix implicit to and from bool type juggling #60890

[W.I.P.] Fix implicit to and from bool type juggling #60890

Uh oh!

Girgias commented Jun 24, 2025 •

edited

Loading

Uh oh!

carsonbot commented Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof commented Jun 24, 2025

Uh oh!

Girgias commented Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

Girgias Jun 24, 2025

Uh oh!

nicolas-grekas Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

stof Jun 24, 2025

Uh oh!

Uh oh!

Uh oh!

[W.I.P.] Fix implicit to and from bool type juggling #60890

Are you sure you want to change the base?

[W.I.P.] Fix implicit to and from bool type juggling #60890

Uh oh!

Conversation

Girgias commented Jun 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

carsonbot commented Jun 24, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

stof commented Jun 24, 2025

Uh oh!

Girgias commented Jun 24, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Girgias commented Jun 24, 2025 •

edited

Loading