Thanks for your pull request, @rainers!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

⚠️⚠️⚠️ Warnings ⚠️⚠️⚠️

• In preparation for migrating from Bugzilla to GitHub Issues, the issue reference syntax has changed. Please add the word "Bugzilla" to issue references. For example, Fix Bugzilla Issue 12345 or Fix Bugzilla 12345.(Reminder: the edit needs to be done in the Git commit message, not the GitHub pull request.)

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "stable + dmd#23263"

@ibuclaw I think I covered your concerns. Still any objections?

I hope you don't need to deal with/break ClassKind.objc stuff.

I hope you don't need to deal with/break ClassKind.objc stuff.

dmd crashes when cross-compiling for macos, but I tried LDC on

void main()
{
    Object obj = new Object;
    auto objc = cast(ObjectiveC)(obj);
    auto nsstring = cast(NSString)(obj);
    auto objc2 = cast(ObjectiveC)(nsstring);
    auto obj2 = cast(Object)(nsstring);
}

extern (Objective-C)
extern interface NSString
{
    import core.attribute : selector;
    const(char)* UTF8String() @selector("UTF8String");
}

extern (Objective-C) class ObjectiveC {}

class classObjC2D : ObjectiveC, NSString
{
    extern (Objective-C)
    override const(char)* UTF8String() { return null; }
}

[casting between extern(C++) and extern(Objective-C) doesn't seem supported.]

LDC 1.42 generates calls to _d_cast for all assignments in main, but with LDC 1.41 objc_opt_isKindOfClass, objc_opt_isKindOfClass, object_getClass, class_conformsToProtocol and _d_interface_cast get called. Is this a regression?

It still seems this PR might indeed change what LDC does for extern(Objective-C) <-> extern(D) casts.
Does LDC actually support a cast from extern(Objective-C) to extern(D)?

I'm inclined to remove the restriction that the dynamic cast is only used for D -> D/C++, but allow any target interface, no matter what linkage.

Is this a regression?

I guess so, although I'm not aware of any filed issues - our extra glue code for that is still in place, but we expect these casts NOT to be lowered to _d_cast (and druntime doesn't use objc_opt_isKindOfClass etc. in that hook or anywhere else).

Edit: Or maybe that was/is just an overshooting implementation in LDC, and relying on the D declarations incl. _d_cast() enough.

Edit2: Some minimal context in https://github.com/ldc-developers/ldc/pull/4777/changes#r1863875351. I guess this was added to support Swift class/interface stubs (so I guess incomplete/opaque bindings on the D side). But I guess this could all be handled in _d_cast in druntime nowadays instead (and then not just LDC-specific).

I'm inclined to remove the restriction that the dynamic cast is only used for D -> D/C++, but allow any target interface, no matter what linkage.

Yeah I think that makes sense.