Matthew Brown

It's not "stricter" behaviour that we're discussing, though — it's sound vs unsound. Unlike other unsound stuff in typecheckers (like array keys being converted to ints in some cases) this...

Psalm also still supports (actually also incorrectly-thought-out) array annotation `array{a: Foo, b: Bar}&array` for unsealed array shapes. That annotation is (correctly) a no-op in PHPStan. For users who encounter this...

Thanks for flagging. I occasionally remember Psalm doesn't warn about this.

> AOT requires a mechanism for traversing an entire codebase to find all classes/files that could be required, and compiling them all. This would also require a mechanism to tell...

@viratyosin yes! But you don't see the second `HH_NOT_A_FIXME[4110]` in the non-full-fidelity parse tree, because it's attached to the trailing trivia of the open bracket (as opposed to the leading...

> you'll just end up with cases where you can't see the comment associated with a closing bracket rather than an opening one. I don't see how that could happen...

As to the `HH_FIXME[8...]` proposal, that's very kind! But I don't think it's a great solution for us — I prefer to use explicit FIXME names rather than force devs...

> And by the way, if you have a working example of your use case that you can share, that would be wonderful ;) Hey - I have a use-case...

Would you mind reset that branch to the upstream head? I think it might disappear.

Hey! The main worry I have here is the increased burden on Psalm’s volunteer maintainers when someone starts using security analysis via this workflow, having not used Psalm before. Psalm’s...