Sanity Check -- Is this valid C++?

[Jonathan Pryor]

I'm in need of a sanity check: is the following code valid C++?

namespace foo {
class NamespaceClass {
};
}

using namespace foo;

class Bug_GlobalFriendDeclaresNamespaceClass {
friend class NamespaceClass;
};

class UseNamespaceClass {
public:
NamespaceClass* GetFoo ();
friend class NamespaceClass;
};

int main ()
{
}

This compiles for me under GCC 3.3.3 (Linux), GCC 3.4.4 (cygwin), GCC
4.0.0 20041026 (Apple build 4061 on a PPC Mac).

This doesn't compile for me under GCC 4.0.1 (Apple build 5250 on an
intel mac). I suspect a compiler bug, but I'd like to make sure the
code itself isn't faulty before blaming the compiler. GCC complains
with the error:

friend.cpp:14: error: ISO C++ forbids declaration of 'NamespaceClass'
with no type
friend.cpp:14: error: expected ';' before '*' token

It seems that the friend declaration within
Bug_GlobalFriendDeclaresNamespaceClass introduces a new NamespaceClass
declaration (presumably at the global scope) which screws up the
UseNamespaceClass use of NamespaceClass.

Any ideas?

Thanks,
- Jon

[Victor Bazarov]

Jonathan Pryor wrote:

I'm in need of a sanity check: is the following code valid C++?

namespace foo {
class NamespaceClass {
};
}

using namespace foo;

class Bug_GlobalFriendDeclaresNamespaceClass {
friend class NamespaceClass;
};

class UseNamespaceClass {
public:
NamespaceClass* GetFoo ();
friend class NamespaceClass;
};

int main ()
{
}

This compiles for me under GCC 3.3.3 (Linux), GCC 3.4.4 (cygwin), GCC
4.0.0 20041026 (Apple build 4061 on a PPC Mac).

This doesn't compile for me under GCC 4.0.1 (Apple build 5250 on an
intel mac). I suspect a compiler bug, but I'd like to make sure the
code itself isn't faulty before blaming the compiler. GCC complains
with the error:

friend.cpp:14: error: ISO C++ forbids declaration of 'NamespaceClass'
with no type
friend.cpp:14: error: expected ';' before '*' token
And which line is 14?
It seems that the friend declaration within
Bug_GlobalFriendDeclaresNamespaceClass introduces a new NamespaceClass
declaration (presumably at the global scope) which screws up the
UseNamespaceClass use of NamespaceClass.
Sounds plausible.
Any ideas?

The simplest way to check with yet another compiler is to go to Comeau
Computing web site: http://www.comeaucomputing.com/tryitout/ and compile
your code there. I did. It compiled. GCC project has its own web site
and its own forums, perhaps you should consider posting there.

V
--
Please remove capital As from my address when replying by mail

[Sumit Rajan]

"Jonathan Pryor" <jo******@vt.edu> wrote in message
news:11**********************@o13g2000cwo.googlegr oups.com...

I'm in need of a sanity check: is the following code valid C++?

namespace foo {
class NamespaceClass {
};
}

using namespace foo;

class Bug_GlobalFriendDeclaresNamespaceClass {
friend class NamespaceClass;
};

class UseNamespaceClass {
public:
NamespaceClass* GetFoo ();
friend class NamespaceClass;
};

int main ()
{
}

This compiles for me under GCC 3.3.3 (Linux), GCC 3.4.4 (cygwin), GCC
4.0.0 20041026 (Apple build 4061 on a PPC Mac).

This doesn't compile for me under GCC 4.0.1 (Apple build 5250 on an
intel mac). I suspect a compiler bug, but I'd like to make sure the
code itself isn't faulty before blaming the compiler. GCC complains
with the error:

friend.cpp:14: error: ISO C++ forbids declaration of 'NamespaceClass'
with no type
friend.cpp:14: error: expected ';' before '*' token

It seems that the friend declaration within
Bug_GlobalFriendDeclaresNamespaceClass introduces a new NamespaceClass
declaration (presumably at the global scope) which screws up the
UseNamespaceClass use of NamespaceClass.

Any ideas?

Compiled it with VC++8.0 and with Comeau C++ 4.3.3. No problems encountered.

Regards,
Sumit.
--
Sumit Rajan <su****@msdc.hcltech.com>