Standard conversion sequences now have a CopyConstructor field, to
cope with the case where a user-defined conversion is actually a copy
construction, and therefore can be compared against other standard
conversion sequences. While I called this a hack before, now I'm
convinced that it's the right way to go.

Compare overloads based on derived-to-base conversions that invoke
copy constructors. 

Suppress user-defined conversions when attempting to call a
user-defined conversion.



git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@58629 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Sema/SemaOverload.h b/lib/Sema/SemaOverload.h
index 2fd6b1f..af3d1b4 100644
--- a/lib/Sema/SemaOverload.h
+++ b/lib/Sema/SemaOverload.h
@@ -18,6 +18,7 @@
 #include "llvm/ADT/SmallVector.h"
 
 namespace clang {
+  class CXXConstructorDecl;
   class FunctionDecl;
 
   /// ImplicitConversionKind - The kind of implicit conversion used to
@@ -114,6 +115,13 @@
     /// is an opaque pointer that can be translated into a QualType.
     void *ToTypePtr;
 
+    /// CopyConstructor - The copy constructor that is used to perform
+    /// this conversion, when the conversion is actually just the
+    /// initialization of an object via copy constructor. Such
+    /// conversions are either identity conversions or derived-to-base
+    /// conversions.
+    CXXConstructorDecl *CopyConstructor;
+
     void setAsIdentityConversion();        
     ImplicitConversionRank getRank() const;
     bool isPointerConversionToBool() const;