lanza added a comment.
> I don't think it'll actually error out at link time: protocol objects get
> emitted eagerly on use, cross-module linking is just a code-size
> optimization. This actually has caused longstanding problems.
But if it's just a forward declaration there's nothing to emit. The above code
compiles just fine as is with just a warning. Here's the result of `clang
protocol.m -lobjc`
proto.m:10:31: warning: cannot find protocol definition for 'NonRuntimeProto'
@interface Implementer : Root<NonRuntimeProto>
^
proto.m:8:11: note: protocol 'NonRuntimeProto' has no definition
@protocol NonRuntimeProto;
^
1 warning generated.
Undefined symbols for architecture x86_64:
"__OBJC_PROTOCOL_$_NonRuntimeProto", referenced from:
__OBJC_CLASS_PROTOCOLS_$_Implementer in proto-bd4a43.o
ld: symbol(s) not found for architecture x86_64
clang-12: error: linker command failed with exit code 1 (use -v to see
invocation)
The protocol definition isn't actually required to compile an implementation.
And if that protocol is declared as `objc_non_runtime_protocol` it won't ever
see one.
Simply requiring that it is annotated accordingly also isn't satisfactory for
the same inheritance problem you mentioned above
__attribute__((objc_non_runtime_protocol))
@protocol SomeProto;
can tell clang not to emit it but won't let clang know if there's a base
protocol that still needs to be emitted. e.g. if we have
@protocol Base
@end
__attribute__((objc_non_runtime_protocol))
@protocol SomeProto <SomeBaseProtol>
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D75574/new/
https://reviews.llvm.org/D75574
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
