[regression] DOMElement::$attributes is considered nullable

[Jean85]

Bug report

Upgrading from 2.1.16 to 2.1.17 facile-it/paraunit#318

I get a regression where DOMElement::$attributes is considered nullable, where it's not. Proof: https://www.php.net/manual/en/class.domnode.php#domnode.props.attributes

The property is nullable for its ancestor (\DOMNode), but not in this case.

Code snippet that reproduces the problem

https://phpstan.org/r/ce94a684-3de4-4a48-9441-b5586c967fac

Expected output

No errors, since it shouldn't be nullable.

Did PHPStan help you today? Did it make you happy in any way?

Totally! So many bug catched early!

[VincentLanglet]

The PR JetBrains/phpstorm-stubs#1741 might be wrong ; do you want to provide a PR to revert it @Jean85 ?

But then, we'll get the always true issue reported by
https://phpstan.org/r/a7bb5a31-9cb5-4f07-8ca7-a3d266ad1107

Related PR phpstan/phpstan-src#3800.

There is maybe a bug to fix first with phpstan-assert-if-true

[Jean85]

Yes I agree, JetBrains/phpstorm-stubs#1741 IS wrong. That attribute is overriding \DOMNode::$attributes, which is correctly marked as nullable: https://github.com/VincentLanglet/phpstorm-stubs/blob/fed4bb7b17ab85f9365a3a77cf9246d1eb5c1b08/dom/dom_c.php#L76-L81

I'll suggest a revert.

[VincentLanglet]

PR on PHPstorm is merged, the PR on PHPStan is ready phpstan/phpstan-src#4029

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.