================
@@ -3363,6 +3363,7 @@ static void encodeTypeForFunctionPointerAuth(const
ASTContext &Ctx,
#include "clang/Basic/RISCVVTypes.def"
llvm_unreachable("not yet implemented");
}
+ break;
----------------
AaronBallman wrote:
My thinking was: if we add the `llvm_unreachable` as the last statement in the
case for `Type::Builtin`, then any case in the nested switch which does not
return will eventually hit that unreachable and be caught as a bug rather than
falling through. If someone adds a new case to the nested switch, this still
happens.
As for the diagnostic, it's off by default (https://godbolt.org/z/PPs993TT9)
but is something we explicitly opt into for the project to help us catch bugs:
https://github.com/llvm/llvm-project/blob/bee2654300a8f524e05dd4cad41411d597246ac0/llvm/cmake/modules/HandleLLVMOptions.cmake#L824
https://github.com/llvm/llvm-project/pull/99763
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
