Re: c++/8633: [3.3 regression] typedefs as base classes

[bangerth at dealii dot org]

Old Synopsis: g++ doesn't resolve typedefs of templated classes properly
New Synopsis: [3.3 regression] typedefs as base classes

State-Changed-From-To: open->analyzed
State-Changed-By: bangerth
State-Changed-When: Tue Nov 19 07:13:10 2002
State-Changed-Why:
    Confirmed. This is a reduced testcase:
    -----------------------
    namespace NS {
      class X {};
      typedef X Y;
    }
    
    struct Base : virtual public NS::Y {
        Base() : NS::Y() {}
    };
    -------------------------
    The messages are
    tmp/g> /home/bangerth/bin/gcc-3.3x-pre/bin/c++ x.cc
    x.cc: In constructor `Base::Base()':
    x.cc:7: error: type `NS::Y' is not a direct or virtual base of `Base'
    
    I am unsure whether the code is illegal: 10.1 states that
    the names of base classes shall be class names, so a
    typedef may not qualify. However, gcc should then tell us
    so at the point of the base class list, not in the
    constructor. What is also suspicious is that the error
    goes away if one makes the base class non-virtual or moves
    X and Y into the global namespace. Something's wrong here.
    
    I can compile the testcase above with 2.95 and 3.2.1pre, so
    assuming it is legal, it is a regression. I thus raise its
    priority. If it is not legal, someone will give an explanation
    of this inconsistent behavior.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=8633