Re: C++, 2.95: Maybe bug with nested template specialization (variant)

[Wolfgang Bangerth <Wolfgang dot Bangerth at IWR dot Uni-Heidelberg dot De>]

> > Line 1, 2 and 4 are fine, but line three is misleading. It goes away when
> > the two 'template <>' are removed. Shouldn't they be needed?
> 
> Yup.  Please submit yet another bug report about it.

I couldn't reproduce the problem without the otherwise buggy lines, so the
message appears only when the code is broken anyway. If you want me to, I
can send a bug report anyway, but for me its not important since one has
to fix the code in any case.

I have another ICE, however. Our whole university is on gcc's spam-filter
list, so I can't post it myself and therefore send it as an attachment to
you. Could you be so kind as to relay this message to gcc-bugs for me?

Many thanks
  Wolfgang

-------------------------------------------------------------------------
Wolfgang Bangerth          email: wolfgang.bangerth@iwr.uni-heidelberg.de
                             www: http://gaia.iwr.uni-heidelberg.de/~wolf

// The following little program ICEs. To get the ICE, the 'typename' keyword
// which I consider necessary here does not matter, one could equally well
// omit it.
//
// configuration:
//   gcc 2.95, sun-solaris-2.6
//
// gcc reports:
//    test03.cc: In function `void f<2>()':
//    test03.cc:25:   instantiated from here
//    test03.cc:22: Internal compiler error.
//
// Many thanks and best regards
//   Wolfgang
//
template <int dim>  struct X {
    template <int N> struct T    {};
};

template <int dim>
void f() {
  typename X<dim>::T<1> t;  
};

template void f<2> ();

//-------------------------------------------------------------------------
//Wolfgang Bangerth          email: wolfgang.bangerth@iwr.uni-heidelberg.de
//                             www: http://gaia.iwr.uni-heidelberg.de/~wolf