orklah

That's strange, it should be recognized here: https://github.com/vimeo/psalm/blob/b4fdc3e326fea865de49827f723a2703bb18617a/src/Psalm/Type/Atomic.php#L270 The slash makes me think it tries to find it as a FQN but it shouldn't go that far...

Though it's easy to list array transformations that leads to a list invalidation, it may be way harder to list and handle every case that could lead to an array...

Interesting. I don't have much experience using phar, would you mind providing a PR for that? It would be cool to have this fixed indeed

It was most likely caused by this change: https://github.com/vimeo/psalm/pull/7727/files#diff-41dfd662b2225a26b01170877d516fe5ef1df0ef8d7703cf8527b50bbcf4bac6R66 My guess is that the compatibility layer where the trait is declared twice is badly handled in Psalm (this is not...

Yeah, I guess we could make sure to explain which mode we're in (taint or analysis) in the message displayed at the beggining. I don't want to remove the option...

@AndrolGenhald can you confirm this works as expected now?

It doesn't show an error anymore

I don't get how the unit test is wrong. You should be able to replace `assertNotSame($d, $c);` by `assert($d !== $c);`. It's a valid thing to check

Oh right, didn't thought about that

Seems great. Let's wait a few days to see if Matt or Bruce have time to chime in and we'll be good to go :) Thanks for the massive work!