Re: [cedet-semantic] Semantic does not correctly show PRIVATE/PROTECTED methods to a class of whom those private/protected methods belong

[David E. <de...@ra...>]

Dixon Ryan writes:
> So please revert any changes you applied because one needs to have the
> class keyword!

Yes, the C++03 standard indeed says this. However, when you
forward-declare the class, like

  class A;
  class B
  {
     friend A;
  };

then this is unambiguous and most compilers will happily accept it. I
think in C++11 this is also now "officially" allowed. I just checked and
GCC accepts this since version 4.7 also for C++03. So I think Eric's
patches are the right thing to do.

-David