PHPStan hangs with trait that returns anonymous class that uses the trait

[ptlis]

Bug report

Please review the sample code below. This appears to be specific to anonymous classes - if you switch out the anonymous class for a 'real' class that uses the trait there is no issue (e.g. https://phpstan.org/r/2fc4cc39-6ec4-4063-b904-7efaed985648).

Code snippet that reproduces the problem

I'm unable to share a 'try' link as the sample code hangs on there too, but this is a small test-case:

<?php declare(strict_types = 1);

trait foo {
	public function getFoo() 
	{
		return new class {
			use foo;
		};
	}	
}

class HelloWorld
{
	use foo;
}

var_dump((new HelloWorld())->getFoo()->getFoo());

And you can verify that this is valid (if bad) code via this link: https://3v4l.org/VpFPP

Expected output

PHPStan should not hang.

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

Yep, I'm very satisfied with what I've seen with PHPStan with my use-cases so far.

The code that triggered this hang in the first place was actually wrong, so in spite of the hanging it still provided me with valuable information.

[herndlm]

We already had such things a couple of times in the past, but this one's kind of special. You can then chain-call ->getFoo() theoretically as often as you want, right? Maybe phpstan should just skip such cases. It's a bit weird 😅 I can check that soon.

[ptlis]

Yeah, in theory you should be able chain any number of invocations of getFoo() - though I doubt there are any real-world uses outside this type of contrived scenario.

[herndlm]

Currently working on this, so I "fixed" the recursion in FileTypeMapper but now I got another one in NodeScopeResolver which is not a surprise I guess :) still looking into it this is "only" a NodeScopeResolver problem

[ondrejmirtes]

Fixed: phpstan/phpstan-src#1303

[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.