From owner-freebsd-hackers  Thu Oct 17 14:21:15 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id OAA00762
          for hackers-outgoing; Thu, 17 Oct 1996 14:21:15 -0700 (PDT)
Received: from rocky.mt.sri.com (rocky.mt.sri.com [206.127.76.100])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id OAA00756
          for <hackers@freebsd.org>; Thu, 17 Oct 1996 14:21:11 -0700 (PDT)
Received: (from nate@localhost) by rocky.mt.sri.com (8.7.5/8.7.3) id PAA20617; Thu, 17 Oct 1996 15:19:51 -0600 (MDT)
Date: Thu, 17 Oct 1996 15:19:51 -0600 (MDT)
Message-Id: <199610172119.PAA20617@rocky.mt.sri.com>
From: Nate Williams <nate@mt.sri.com>
To: hackers@freebsd.org
Subject: C++ question
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Any C++ gurus out here?  If so, here's a small snippet of a program that
I am having some questions on.

------------cut here---------------
class foo
{
  public:
    foo() {};

  private:
    struct A {
        int bigint;
    };

    struct B {
        A   *ptr;      // Sun's C++ compiler doesn't like this
    };

    A myints;

    A *getint(void);

};

#if __GNUC__
foo::A *               // G++ requires this, but Sun's compiler gets lost
#else
A *                    // I prefer G++'s syntax above, but can't for Sun
#endif

foo::getint(void)
{
    return ( &myints );
}
------------cut here----------------

The above snippet of code compiles fine with g++ and *all* warnings
turned on, but gives the following warning with Sun's CC compiler
(latest released version).

fly:/tmp % CC +w -c grpdgram.cc
"grpdgram.cc", line 12: Warning (Anachronism): foo::A is not accessible from foo::B.

It's only a warning, but according to the documentation I'm doing
something that won't be allowed in future C++ releases. 

All I'm trying to do is define a structure, and then using the resulting
definition in another structure.  (It's used for *simple* list
processing in the real code, and I have a place-holder for multiple
lists in another structure).  I know how to get rid of the warnings with
the particular application (be more 'object oriented';), but I would
think using two structures would be allowed in C++.

Any clues?

Finally, I find it interesting that G++ and Sun's CC compiler have
completely different function declaration syntax when you return
structures defined inside the function.



Nate