[Bug c++/48010] typedef redefinitions are allowed if the redefined type is a nested class type

["fabien at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48010

fabien at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|fabien at gcc dot gnu.org   |unassigned at gcc dot
                   |                            |gnu.org

--- Comment #3 from fabien at gcc dot gnu.org 2011-03-16 22:37:56 UTC ---


Index: name-lookup.c
===================================================================
--- name-lookup.c    (revision 170721)
+++ name-lookup.c    (working copy)
@@ -459,7 +459,8 @@ supplement_binding (cxx_binding *binding
            && DECL_ANTICIPATED (bval)
            && !DECL_HIDDEN_FRIEND_P (bval)))
     binding->value = decl;
-  else if (TREE_CODE (bval) == TYPE_DECL && DECL_ARTIFICIAL (bval))
+  else if (TREE_CODE (bval) == TYPE_DECL && DECL_ARTIFICIAL (bval)
+       && TREE_CODE (decl) != TYPE_DECL)
     {
       /* The old binding was a type name.  It was placed in
      VALUE field because it was thought, at the point it was

I've tried the patch above, but it raises another (complicated) issue.
The below reduced testcase no longer compile. While instanciating A<int>, two
decl of 'b' reach push_class_level_binding, and with the patch above, the
second call to supplement_binding no longer succeeds.

template <class>
struct A
{
    typedef struct {} b;
};

template class A<int>;

I don't know if we are wrongly instanciating something or if we need to treat
specially a typedef to an anonymous struct in supplement_binding.Unless someone
gives me hint, I give up.