SE-0384: Importing Forward Declared Objective-C Interfaces and Protocols

[NuriAmari]

@edudnyk

I would argue that the whole ClangImporter is a deferred follow-up (example ), which never actually achieved a feature parity with the compiler. It more often fails than succeeds to import mixed modules that were successfully compiled. Adding more to the already existing debt and separation of the language support does not seem like a proper way of addressing the problem.

Regardless of one's stance on if the existing debt should be increased via this proposal, I agree that the ClangImporter and its interactions with the rest of the system have some existing fundamental issues that need addressing before compiler - debugger parity is achievable.

With respect to the issue you've linked, I'm aware of this issue and have recently made a rough attempt at fixing it (the compiler part of it): [ClangImporter] Swift Implemented ObjC Forward Declarations by NuriAmari · Pull Request #62983 · apple/swift · GitHub. Though decidedly related, unless the community feels otherwise I'd rather not stray too far with this proposal's discussion. A solution to that problem relies upon a resolution of the issues raised in this discussion anyhow.

[allevato (Tony Allevato)]

This proposal has been accepted. Thanks to all who participated in the review!