[Bug c++/13451] New: [3.4 regression] Wrong error message with qualified names for member declarations

["giovannibajo at libero dot it" <gcc-bugzilla at gcc dot gnu dot org>]

Consider this:

template <class T>
struct A
{
  struct B;
  struct A::B { };    // #1
  void A::foo(void);  // #2
};

>From the standard it's not clear whether #1 and #2 should be accepted or 
rejected. We used to accept both since 2.95, but on mainline we now get a hard 
error on #1:

test.cc:5: error: too few template-parameter-lists

Looks like the injected class name confuses the parser. I rate this a 
regression because:

1) It's an unwanted change of behaviour.
2) Even if we now want to reject #1 for some reason, we need to do that with a 
proper error message. 


Then, we could discuss what to do with these. EDG rejects both saying that 
member declarations can't be qualified. For not, I assume we want to preserve 
our historical behaviour, so this is reject-valid.

-- 
           Summary: [3.4 regression] Wrong error message with qualified
                    names for member declarations
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Keywords: rejects-valid
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: giovannibajo at libero dot it
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13451