Question

Here my code snippet:

class Request
{
public:
 Request(void);
………..
}

Request::Request(void)
{
 qDebug()<<"Request: "<<"Hello World";
}


class LoginRequest :public Request
{
public:
 LoginRequest(void);
 LoginRequest(QDomDocument);
……………
}

LoginRequest::LoginRequest(void)
{
 qDebug()<<"LoginRequest: "<<"Hello World";
 requestType=LOGIN;
 requestId=-1;   
}

LoginRequest::LoginRequest(QDomDocument doc){
 qDebug()<<"LoginRequest: "<<"Hello World with QDomDocument";
 LoginRequest::LoginRequest();       
 xmlDoc_=doc;         
}

When call constructor of Overrided LoginRequest

LoginRequest *test=new LoginRequest(doc);

I came up with this result:

Request:  Hello World
LoginRequest:  Hello World with QDomDocument
Request:  Hello World
LoginRequest:  Hello World

Obviously both constructor of LoginRequest called REquest constructor.

Is there any way to cape with this situation?

I can construct another function that does the job I want to do and have both constructors call that function. But I wonder is there any solution?

Edit:
http://www.parashift.com/c++-faq-lite/ctors.html#faq-10.3

Answer 1

The code is not doing what you probably think it is doing. The line:

 LoginRequest::LoginRequest();     

constructs a temporary object which is immediately destroyed. as others have suggested, you can put duplicate code in a private function, but this has a lot of issues – specifically, such a function can only perform assignment, not initialisation, and many classes do not support assignment. A somewhat better solution is to implement a single constructor with a default argument:

class LoginRequest {
   ....
   LoginRequest( QDomDocument d = DefaultDoc() );
};