[clang][modules] Don't hard code [no_undeclared_includes] for the Darwin module

[iana]

Summary

The Darwin module has specified [no_undeclared_includes] for at least five years now, there's no need to hard code it in the compiler.

[ributzka]

Seems straightforward. LGTM.

[Bigcheese]

I'm fine with this change, but do we actually have a backwards compatibility policy anywhere in Clang? Would be good to know what range of SDKs a compiler release is expected to support.

[iana]

In D132971#3759538, @Bigcheese wrote:

I'm fine with this change, but do we actually have a backwards compatibility policy anywhere in Clang? Would be good to know what range of SDKs a compiler release is expected to support.

I don't know about backwards compatibility, but I found the Darwin module added the attribute in 2017 macOS 10.13/iOS 11.0. As far as I can tell from the original change, adding the attribute didn't actually fix anything, it was just used to detect cycles between libc++ and Darwin, so I think removing the hard code wouldn't even break older OSes.

clang/lib/Lex/ModuleMap.cpp

ActiveModule =
             .first;
   }
 
   ActiveModule->DefinitionLoc = ModuleNameLoc;
   if (Attrs.IsSystem || IsSystem)
     ActiveModule->IsSystem = true;
   if (Attrs.IsExternC)
     ActiveModule->IsExternC = true;
-  if (Attrs.NoUndeclaredIncludes ||
-      (!ActiveModule->Parent && ModuleName == "Darwin"))
+  if (Attrs.NoUndeclaredIncludes)
     ActiveModule->NoUndeclaredIncludes = true;
   ActiveModule->Directory = Directory;
 
   StringRef MapFileName(ModuleMapFile->getName());
   if (MapFileName.endswith("module.private.modulemap") ||
       MapFileName.endswith("module_private.map")) {
     ActiveModule->ModuleMapIsPrivate = true;
   }