How to force overloaded call in derived classes?
{ please note that some lines in the code below were wrapped making the
code ill-formed. reassemble the lines to compile the code.
to all: please format your code to follow the 70 character guideline,
break up long lines so that they are still OK when copied and pasted
into a text editor for compilation. thanks! -mod }
First of all thanks for your patience and good pointers
for Visitor and Curiously Recurring Template patterns last time :)
Now another problem of mine.
THE CODE
#include <iostream>
#include <vector>
struct CMessage;
struct CTextMessage;
struct CServerMessage;
struct CMessageInterface;
struct CMessage;
class CMessageProcessor
{
public:
virtual void ProcessMessage(CMessage& m) { std::cout << "Processing
any
message" << std::endl; };
// virtual void ProcessMessage(CTextMessage& a) { std::cout <<
"Processing CTextMessage" << std::endl; }; // (1)
// virtual void ProcessMessage(CServerMessage& c) { std::cout <<
"Processing CServerMessage" << std::endl; };
};
class CAnotherProcessor: public virtual CMessageProcessor
{
public:
virtual void ProcessMessage(CTextMessage& a) { std::cout <<
"Processing
CTextMessage in Another" << std::endl; };
};
class CDerivedProcessor: public CAnotherProcessor
{
public:
virtual void ProcessMessage(CTextMessage& a) { std::cout <<
"Processing
CTextMessage in Derived" << std::endl; };
};
struct CMessage
{
virtual void Process(CMessageProcessor& p) = 0;
virtual ~CMessage() {};
};
template<typename Self>
struct CMessageBase: public CMessage
{
virtual void Process(CMessageProcessor& p) {
p.ProcessMessage(*static_cast<Self*>(this)); }; // (2)
};
struct CMessage: public CMessageBase<CMessage>
{
};
struct CTextMessage: public CMessageBase<CTextMessage>
{
};
struct CServerMessage: public CMessageBase<CServerMessage>
{
};
int main()
{
CDerivedProcessor p;
CAnotherProcessor ap;
CMessageProcessor* pP = &p;
CMessageProcessor* pAP = ≈
std::vector<CMessage*> v;
v.push_back(new CMessage());
v.push_back(new CTextMessage());
v.push_back(new CServerMessage());
for (int i = 0; i < 3; ++i)
{
CMessage& rMsg = *v.at(i);
rMsg.Process(*pP); // (3)
rMsg.Process(*pAP);
//v.at(i)->Process(*pP);
//v.at(i)->Process(*pAP);
}
while (!v.empty())
{
delete v.back();
v.pop_back();
}
return 0;
};
THE PROBLEM
An overloaded ProcessMessage() for CTextMessage in CAnotherProcessor and
CDerivedProcessor is called only when appropriate ProcessMessage is
defined in CMessageProcessor.
Seems obvious since at (3) the virtual function (1) of that definition
is unknown and virtuality doesn't play.
However I missed this since from the beginning I've used
CMessageProcessor with all these functions uncommented.
THE QUESTION
How can I make it that at (3) the call of (2) is done to the function I
want?
Ie: when rMsg at (3) is really CTextMessage and
CDerivedProcessor::ProcessMessage(CTextMessage) is called
instead of CMessageProcessor::ProcessMessage(CMessage)?
THE NEED
I would like to have several custom message processors that are
able to process custom messages only by defining an overloaded
ProcessMessage(CCustomMessageType) and calling Process() on
CCustomMessageType it will call appropriate ProcessMessage regardles
of what is defined in CMessageProcessor.
So: extending a CCustomProcessor to process new message types
should require only defining a new ProcessMessage(CCustomMessageType)
- at least this is what I would like to have.
Thank you
-Marcin
--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]