Nice trick or illegal construct?

[rmilh]

As you can see in this example the coder did't think it is necessary
to have an instance of a Derived to be able to set Base member prot.
Is it legal?

class Base
{
protected:

int prot;
};

class Derived : public Base
{
public:

using Base:rot;
};

int main()
{
Base base;
//instance of derived not needed to access protected member?
static_cast<Derived&>(base).prot = 1;
return 0;
}

Robert Milharcic

 

[Ivan Vecerina]

As you can see in this example the coder did't think it is necessary
to have an instance of a Derived to be able to set Base member prot.
Is it legal? ....
Base base;
//instance of derived not needed to access protected member?
static_cast<Derived&>(base).prot = 1;

Not a good idea.
Formally, casting a base class instance as a derived type triggers
undefined behavior (even though the above code is likely to work
"as expected" on many platforms).


Cheers, Ivan

 

[Tom|420]

static_cast<Derived&>(base).prot = 1;

dynamic_cast is more appropriate for classes

 

[Andre Caldas]

Hello!

Is it legal?

class Base *snip*

class Derived : public Base *snip*

int main()
{
Base base;
//instance of derived not needed to access protected member?
static_cast<Derived&>(base).prot = 1;
return 0;
}

This only works because Base starts at the "begining" of Derived. If you
had:
class Derived: public A, public Base, public C
{
...
}

Then I think it would not work.

Andre Caldas.

 

[Clark S. Cox III]

As you can see in this example the coder did't think it is necessary
to have an instance of a Derived to be able to set Base member prot.
Is it legal?
No.

class Base;
class Derived : public Base;

Base base;
static_cast<Derived&>(base)

Your cast is undefined behavior.