[Bug c++/24671] New: [4.0/4.1 regression] ICE with zero-sized arrays

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

The following testcase causes an ICE:

==============================================
template<typename> struct A
{
    typedef int X;
    static const int i = 0;
};

template<typename> struct B
{
    typedef typename A<char[A<B>::i]>::X Y;
    template<typename T> B(T, Y);
};

B<int> b(0,0);
==============================================

dPR19208B.cc:10: internal compiler error: in push_access_scope, at cp/pt.c:176
Please submit a full bug report, [etc.]

The testcase crashed since at least gcc 2.95.3, but was fixed with the
patch for PR19208. The problem reappeared on the 3.4 branch and mainline
with Josh's patch for PR 19989.

I think the code should be accepted at least with -fpermissive,
but rejected with -pedantic.


-- 
           Summary: [4.0/4.1 regression]  ICE with zero-sized arrays
           Product: gcc
           Version: 4.0.3
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code, ice-on-invalid-code, monitored
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: reichelt at gcc dot gnu dot org
 BugsThisDependsOn: 19208,19989


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24671