Randy Schütt

@ahmadnassri Could you take a look at it? I would very much like to have this feature for php.

The same seems to be true for php / composer. Although phpunit is declared (as it should be) under `require-dev`, it is identified as a production type.

@Azakur4 It would be helpful if at least the type/class of the function would be there. I could help you if you want. You could use my [PHP-Parser](https://github.com/Dgame/PHP-Parser) to extract...

Have you tried another mode, like `Solid`, `Shaded` and `Blended`? It can be different depending on the mode. Since you say that you have never experienced this behaviour anywhere else,...

@staabm What led you to the conclusion that this could be the case? Could you share your test code with me so I can reproduce it?

@staabm I added my own custom rule to test the problem in my project. I found out that reducing the type (if it is a union type) will solve all...

Reduced example: https://phpstan.org/r/aa3d6eea-69cb-421d-8b65-429fb0f30d9b > I don't get it - what else can it be besides A and B? Not sure, but apparently it now thinks that the second match-arm is...

How? Could be both, either A **or** B, so no error should occur: https://3v4l.org/9Q7G9 As you see, the second match-arm is not true if we call it with `test(new A());`

I use `true` as a condition and either is `$t instanceof A` **or** `$t instanceof B`, but never both. So all correct, there should be no error. So I am...

BTW, just tested, neither psalm nor phan report an error here. Normally I think phpstan is more accurate, but in this particular case I'm a bit doubtful, no offense of...