Reference to an undefined assignment operator of a base class in a derived class.

[Vladimir Grigoriev]

Let consider the following code

struct A
{
A( int i = 0 ): x( i ) {}
int x;
};

struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

As you can see in the base class there is no an assignment operator defined
explicitly. However in the derived class there is explicit reference to the
assignment operator of the base class.

The problem is the following. When I use some non-Microsoft compiler it
issues an error for this code. However if I will include an assignment
operator in the base class explicitly the error dissapears.
The Microsoft compiler VC++ 2005 EE compiles the code without any error.
Is it a bug of the non-Microsoft compiler? May I specify explicitly an
assignment operator in a derived class when in a base class it is absent and
defined by default by a compiler? What does C++ standard say about this?
Also as you can see as for a copy constructor there is not such problem when
the code compiled with the non-Micosoft compiler.

Vladimir Grigoriev

[Victor Bazarov]

Post by Vladimir Grigoriev
Let consider the following code
struct A
{
A( int i = 0 ): x( i ) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );

Try

static_cast<A&>(*this) = rhs;

Does your "some non-Microsoft" compiler complain? If so, it's broken.
Both forms are pretty much idiomatic.

Post by Vladimir Grigoriev
return ( *this );
}
int &value;
};
As you can see in the base class there is no an assignment operator defined
explicitly. However in the derived class there is explicit reference to the
assignment operator of the base class.
The problem is the following. When I use some non-Microsoft compiler it
issues an error for this code. However if I will include an assignment
operator in the base class explicitly the error dissapears.
The Microsoft compiler VC++ 2005 EE compiles the code without any error.
Is it a bug of the non-Microsoft compiler? May I specify explicitly an
assignment operator in a derived class when in a base class it is absent and
defined by default by a compiler? What does C++ standard say about this?
Also as you can see as for a copy constructor there is not such problem when
the code compiled with the non-Micosoft compiler.

You can always use Comeau compiler online tryout (find it on their
website: www.comeaucomputing.com) to verify some constructs. Comeau is
considered as close to Standard-compliant as one can get, and they are
very good at responding to questions if you doubt the correctness of
some behaviour of the compiler.

V

[Igor Tandetnik]

Post by Vladimir Grigoriev
Let consider the following code
struct A
{
A( int i = 0 ): x( i ) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};
As you can see in the base class there is no an assignment operator defined
explicitly. However in the derived class there is explicit reference to the
assignment operator of the base class.
The problem is the following. When I use some non-Microsoft compiler it
issues an error for this code. However if I will include an assignment
operator in the base class explicitly the error dissapears.
The Microsoft compiler VC++ 2005 EE compiles the code without any error.
Is it a bug of the non-Microsoft compiler?

Yes. The code is correct as written.

Post by Vladimir Grigoriev
May I specify explicitly an
assignment operator in a derived class when in a base class it is absent and
defined by default by a compiler?

Yes.

Post by Vladimir Grigoriev
What does C++ standard say about this?

Nothing in particular, beyond the fact that

12.8p10 ...An implicitly-declared copy assignment operator is an inline public member of its class...

There's just nothing special about an implicitly declared assignment operator, as compared to its explicitly declared brethren. It's a class member function like any other.

Post by Vladimir Grigoriev
Also as you can see as for a copy constructor there is not such problem when
the code compiled with the non-Micosoft compiler.

Well, I can't see that - I don't have a copy of that mysterious "non-Microsoft compiler". But I trust you at your word.

[Vladimir Grigoriev]

"Igor Tandetnik" <***@mvps.org> wrote in message news:%237mqoA$***@TK2MSFTNGP06.phx.gbl...

Well, I can't see that - I don't have a copy of that mysterious
"non-Microsoft compiler". But I trust you at your word.
--
With best wishes,
Igor Tandetnik

This mysterious "non_Microsoft compiler" is the old Borland C++ Builder 5.0
compiler.

By the way it does not have the bug of Microsoft C++ STL vector
implementation as Visual C++ 2005 does have which I pointed out early.:) So
the problem is haw to join these compilers together such a way that at least
a simple C++ program exept "Hello world" would be compiled according to the
C++ standard.:)

Vladimir Grigoriev
.

With sufficient thrust, pigs fly just fine. However, this is not necessarily
a good idea. It is hard to be sure where they are going to land, and it
could be dangerous sitting under them as they fly overhead. -- RFC 1925

[Darko Miletic]

Post by Vladimir Grigoriev
Let consider the following code
struct A
{
A( int i = 0 ): x( i ) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

Compiling that code in BDS 2006 (bcc32 5.8 ) I get the following error:

[C++ Error] Unit1.cpp(17): E2451 Undefined symbol '='
Unit1.cpp(10): class B
Unit1.cpp(21): decision to instantiate: B & B::operator =(const B &)
--- Resetting parser context for instantiation...
Unit1.cpp(16): parsing: B & B::operator =(const B &)

However adding definition of destructor in class A and B problem goes away:


struct A {
A( int i = 0 ): x( i ) {}
virtual ~A(void) {}
int x;
};

struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
virtual ~B(void) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

[Victor Bazarov]

Post by Darko Miletic

Post by Vladimir Grigoriev
Let consider the following code
struct A
{
A( int i = 0 ): x( i ) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

[C++ Error] Unit1.cpp(17): E2451 Undefined symbol '='
Unit1.cpp(10): class B
Unit1.cpp(21): decision to instantiate: B & B::operator =(const B &)
--- Resetting parser context for instantiation...
Unit1.cpp(16): parsing: B & B::operator =(const B &)
struct A {
A( int i = 0 ): x( i ) {}
virtual ~A(void) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
virtual ~B(void) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

Wow... No surprise nobody uses that compiler anymore.

V

[Vladimir Grigoriev]

I think that this effect is relative to the BORLAND compatible option in the
project properties. If to select the ANSI compatible option the problem
dissapears. Maybe it is relative to the Borland VCL library that is used in
C++ and in Pascal simultaneously.

Vladimir Grigoriev

Post by Victor Bazarov

Post by Darko Miletic

Post by Vladimir Grigoriev
Let consider the following code
struct A
{
A( int i = 0 ): x( i ) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

[C++ Error] Unit1.cpp(17): E2451 Undefined symbol '='
Unit1.cpp(10): class B
Unit1.cpp(21): decision to instantiate: B & B::operator =(const B &)
--- Resetting parser context for instantiation...
Unit1.cpp(16): parsing: B & B::operator =(const B &)
struct A {
A( int i = 0 ): x( i ) {}
virtual ~A(void) {}
int x;
};
struct B: public A
{
B( int i = 0 ): A( i ), value( x ) {}
B( const B &rhs ): A( rhs ), value( x ) {}
virtual ~B(void) {}
B & operator =( const B &rhs )
{
A::operator =( rhs );
return ( *this );
}
int &value;
};

Wow... No surprise nobody uses that compiler anymore.
V
--
Please remove capital 'A's when replying by e-mail
I do not respond to top-posted replies, please don't ask