Trapping messages in MFC - Whats the difference?

Question

I was just wondering what (if any) the difference was between the following two message traps in MFC for the function, OnSize(..).

1 - Via Message map:

BEGIN_MESSAGE_MAP(CClassWnd, CBaseClassWnd)
...
    ON_WM_SIZE()
..
END_MESSAGE_MAP()

2 - Via afx_message:

afx_msg type OnSize(...);

They seem to be used interchangeably, which one should be used or does it depend on other factors?

Answer 1

Both parts are necessary to add a message handler to a class. The message map should be declared inside your class, together with declarations for any message handler functions (e.g, OnSize).

class CClassWnd : public CBaseClassWnd {
    ...
    afx_msg void OnSize(UINT nType, int cx, int cy);
    DECLARE_MESSAGE_MAP
};

afx_msg is just an empty placeholder macro - it doesn't actually do anything, but is always included by convention.

The message map is then defined in the class's .cpp file:

BEGIN_MESSAGE_MAP(CClassWnd, CBaseClassWnd)
    ON_WM_SIZE()
END_MESSAGE_MAP()

These macros generate a lookup table for the class which allows messages received by the window to be dispatched to the corresponding handler functions. The ON_WM_SIZE macro allows the wParam and lParam message parameters in the WM_SIZE message to be decoded into more meaningful values for the message handler function (nType, cx, and cy in this case). MFC provides macros for most window messages (WM_LBUTTONDOWN, WM_DESTROY, etc).

You can find more information on how message maps work in MFC here on MSDN.