This is the mail archive of the gcc-prs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: c++/5094: partial specialisation cannot be friend??


The following reply was made to PR c++/5094; it has been noted by GNATS.

From: Pal Benko <benko@sztaki.hu>
To: rodrigc@gcc.gnu.org
Cc: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
 nobody@gcc.gnu.org
Subject: Re: c++/5094: partial specialisation cannot be friend??
Date: Fri, 14 Dec 2001 09:08:51 +0100

 Thank you for the quick answer.
 
 >     This is not a bug in gcc.  
 >     This is not legal code.  When I compile with
 >     the Digital Unix C++ compiler, I get the
 >     following compilation errors:
 >     
 >     cxx: Error: specfriend.cc, line 67: "c" has already been declared in =
 the
 >               current scope
 >         friend class c;
 >     -----------------^
 
 Is
 
 template <>
 struct c<int>
 {
   class inner
   {
     friend class c<int>;
   };
 
   template <typename T>
   friend class d<c<T> >;
 };
 
 correct?  gcc accepts it, but it also accepted the previous version.
 
 >     cxx: Error: specfriend.cc, line 71: class "c" may not have a template
 >               argument list
 >       friend class d<c<T> >;
 >     -----------------^
 >     cxx: Error: specfriend.cc, line 71: class "c" may not have a template
 >               argument list
 >       friend class d<c<T> >;
 >     -----------------^
 >     cxx: Error: specfriend.cc, line 71: invalid partial specialization --
 >     class
 >               "d<<error-type>>" is already fully specialized
 >       friend class d<c<T> >;
 >     ---------------^
 
 OK, Martin Sebor gave the exact reference from the standard.
 
 >     cxx: Error: specfriend.cc, line 83: class "c" may not have a template
 >               argument list
 >       friend class c<d>;
 >     ---------------^
 
 again: is
 
 template <>
 struct d<int>
 {
   template <int I>
   class inner
   {
     friend class d;
   };
 
   friend class c<d<int> >;
 };
 
 correct?  gcc accepts, as previously.
 
 
 And there are three error messages of gcc not reproduced by the
 Digital Unix C++ in lines 42, 55 and 59 of the original bugreport.  A
 simplified testcase:
 
 struct b
 {
   template <int I>
   class inner
   {};
 };
 
 template <typename T>
 struct c {};
 
 template <typename T>
 struct c<T*>
 {
   class inner
   {
     friend class c;
   };
 };
 
 template <typename T>
 struct c<c<T> >
 {
   template <int I>
   class inner
   {
     friend class c;
   };
 
   template <int I>
   friend class b::inner;
 };
 
 
 Thanks again,
 
 Benko Pal=
 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]