[Bug c++/15075] New: "using namespace" fails in part

[Bug c++/15075] New: "using namespace" fails in part

[gcc-bugzilla at gcc dot gnu dot org]

"using namespace Test;" fails to tell the compiler that a friendly
function definition is within Test.  "namespace Test {...}" works fine.

Environment:
System: SunOS bioquad 5.9 Generic_112233-11 sun4u sparc SUNW,Sun-Fire-V440
Architecture: sun4

	
host: sparc-sun-solaris2.9
build: sparc-sun-solaris2.9
target: sparc-sun-solaris2.9
configured with: ../../../distribution/gcc-3.3.2/configure --prefix=/local/bioproj/people/lnewberg/install/sparc-sun-solaris2.9/gcc-3.3.2 --with-ld=/usr/ccs/bin/ld --with-as=/usr/ccs/bin/as --disable-nls --enable-languages=c,c++,objc

How-To-Repeat:
	Compile with g++ -c test2.cc
--- Begin test2.cc ---
// This file does not compile under g++.  I think it should.  See
// below.

namespace Test
{
    class A
    {
    private:
	class B
	{
	public:
	    int _b;
	};
	friend bool works1(B const &);
	friend bool works2(B const &);
	friend bool fails(B const &);
    };
} // End of namespace Test

namespace Test
{
    bool works1(A::B const &lhs)
    {
	return true;
    }
} // End of namespace Test

using namespace Test;

// Despite the "using namespace Test" directive, the function fails()
// must be declared as "Test::fails" rather than simply "fails" or the
// compiler complains that class Test::A::B is private in this
// context.
// 
// Note that the "using" directive does work to eliminate the need to
// use the explicit Test::A::B instead of the shorter A::B in the
// definition of fails() below.
// 
// Contrastingly, the functions works1(), above, and works2(), below,
// compile without error.

bool Test::works2(A::B const &)
{
    return true;
}

bool fails(A::B const &)
{
    return true;
}
--- End test2.cc ---
------- Additional Comments From lnewberg at wadsworth dot org  2004-04-22 16:55 -------
Fix:
	Use "namespace Test {...}" instead.

-- 
           Summary: "using namespace" fails in part
           Product: gcc
           Version: 3.3.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: lnewberg at wadsworth dot org
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: sparc-sun-solaris2.9
  GCC host triplet: sparc-sun-solaris2.9
GCC target triplet: sparc-sun-solaris2.9


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[lnewberg at wadsworth dot org]

------- Additional Comments From lnewberg at wadsworth dot org  2004-04-22 17:01 -------
Created an attachment (id=6138)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6138&action=view)
G++ source file


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[lnewberg at wadsworth dot org]

------- Additional Comments From lnewberg at wadsworth dot org  2004-04-22 17:02 -------
Created an attachment (id=6139)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6139&action=view)
.ii file for the previous .cc file.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-22 17:12 -------
Actually I believe that works2 should be rejected just as fail is rejected.

namespace Test
{
    class A
    {
        class B
        {
        };
        friend bool works2(B const &);
    };
}
using namespace Test;
bool Test::works2(A::B const &)
{
    return true;
}

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |accepts-invalid


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-04-22 17:28 -------
Andrew, your code is well-formed. In the friend declaration, the unqualified 
name "works2" is looked up and nothing is found. So a name is injected within 
the most enclosing namespace, which is Test. Basically, the friend function 
acts also as a predeclaration for Test::works2().

This is a bug, but I don't have time to confirm/regression-check it now.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-04-22 17:50 -------
Actually, I believe that gcc is right rejecting the code. This is how it 
fails: 
----------- 
namespace Test { 
  class A { 
    private: 
      class B {}; 
      friend void fails(B const &); 
  }; 
} 
 
using namespace Test; 
 
void fails(A::B const &) {} 
-------------- 
 
So we agree that the friend declaration injects the name into namespace Test. 
Which we then have again in the using thing. And then comes the definition 
of fail, but I believe the compiler is correct in assuming that this  
definition has to be in the global namespace -- the using directive only 
gets active once we _look up_ names, not when we declare/define new names. 
So clearly the definition of fails() does not define the predeclared/injected 
Test::fails function, and the compiler is correct in rejecting it due to 
an access violation: 
 
g/x> /home/bangerth/bin/gcc-3.4-pre/bin/c++ -c x.cc 
x.cc: In function `void fails(const Test::A::B&)': 
x.cc:4: error: `class Test::A::B' is private 
x.cc:11: error: within this context 
 
Giovanni, what was your reasoning that differed from this here? 
 
W. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|accepts-invalid             |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

[Bug c++/15075] "using namespace" fails in part

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-04-23 02:42 -------
Yes ok, I didn't properly look at the bug report. I thought GCC was behaving in 
a different way.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075

Re: [Bug c++/15075] "using namespace" fails in part

[Gabriel Dos Reis]

"bangerth at dealii dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| definition has to be in the global namespace -- the using directive only 
| gets active once we _look up_ names, not when we declare/define new names. 

Your analysis is correct.

-- Gaby

[Bug c++/15075] "using namespace" fails in part

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-04-24 20:45 -------
Subject: Re:  "using namespace" fails in part

"bangerth at dealii dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| definition has to be in the global namespace -- the using directive only 
| gets active once we _look up_ names, not when we declare/define new names. 

Your analysis is correct.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15075