The 8.4 CI failure seems to be that the stubs don't actually understand what a covariant template is?

Hey folks, thanks for the PR! Some of the tests are failing at the moment, due to not supported processing of @template T as Type notation. Right now tests just take template name and convert it later to object during comparison with signature. So the tests should be updated to store now only template name, but also template types (\StubTests\Model\PHPDocElement::collectTags) and use an appropriate type later during comparison with function signature (\StubTests\StubsTypeHintsTest::testClassesMethodsSignatureTypeHintsConformPhpDocInMethods). any chance you could take a look and try to fix them? It’ll help get the PR merged faster. If not, no worries, we’ll sort it out on our side later, just with a bit of a delay.

I will try to fix the tests, just wasn't sure what the approach to fix them should be. :)

@Girgias Thank you for fixing tests! All the changes looks good. Could you please take a look at inline comments and fix related items. After that PR can be merged.

@isfedorov, did you forget to submit the inline comments? I can't see any.

@isfedorov, did you forget to submit the inline comments? I can't see any.

🤦 Indeed! Sorry, published review now.

I don't really understand why there are now PGSQL failures on the 8.4 CI :(

@Girgias I believe the failures are not related to your changes but related to the updated PHP 8.4 docker image.

1) StubTests\BaseConstantsTest::testConstants with data set "constant \LIBXML_NO_XXE" ('\LIBXML_NO_XXE')
Missing constant: const \LIBXML_NO_XXE = 8388608

2) StubTests\BaseConstantsTest::testConstants with data set "constant \PGSQL_TUPLES_CHUNK" ('\PGSQL_TUPLES_CHUNK')
Missing constant: const \PGSQL_TUPLES_CHUNK = 12

3) StubTests\BaseFunctionsTest::testFunctionsExist with data set "function \pg_set_chunked_rows_size" ('\pg_set_chunked_rows_size')
Missing function: function \pg_set_chunked_rows_size(Pgsql\Connection $connection, int $size){}

Tests are not bound to a strict version of PHP and use just latest minor updates of major versions so I suppose the related docker image was just updated at some moment in between your commits and new entities were added

@Girgias Could you please revert changes in files under .idea folder? Those are local specific changes and shouldn't be stored in VCS.

@Girgias Could you please revert changes in files under .idea folder? Those are local specific changes and shouldn't be stored in VCS.

Should the folder be added to .gitignore?

@Girgias It's already partially added. I wouldn't add the whole directory since some of its files may be good to share via VCS and also some of the files can contain both, content that does make sense to share and content that doesn't so every single case should be considered separately. Thank you for the fix!