We don't actually need to check the implicit object argument's

conversion a second time for a conversion candidate (with the real
acting context), because the only problems we would find are access or
ambiguity issues that won't be diagnosed until we pick this
candidate. Add a test case to prove it to myself.



git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@111526 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 0 insertions, 15 deletions

diff --git a/lib/Sema/SemaOverload.cpp b/lib/Sema/SemaOverload.cpp
index f8f04defb1..30979b78d3 100644
--- a/lib/Sema/SemaOverload.cpp
+++ b/lib/Sema/SemaOverload.cpp
@@ -3767,21 +3767,6 @@ Sema::AddConversionCandidate(CXXConversionDecl *Conversion,
     return;
   }
 
-  // Make sure that the actual object argument initialization will work, when
-  // it comes down to it. This takes into account the actual acting context.
-  if (ConversionContext->getCanonicalDecl()
-                                        != ActingContext->getCanonicalDecl()) {
-    ImplicitConversionSequence ObjectConvertICS
-      = TryObjectArgumentInitialization(From->getType(), Conversion, 
-                                        ActingContext);
-    if (ObjectConvertICS.isBad()) {
-      Candidate.Viable = false;
-      Candidate.FailureKind = ovl_fail_bad_conversion;
-      Candidate.Conversions[0] = ObjectConvertICS;
-      return;
-    }
-  }
-  
   // We won't go through a user-define type conversion function to convert a 
   // derived to base as such conversions are given Conversion Rank. They only
   // go through a copy constructor. 13.3.3.1.2-p4 [over.ics.user]