Skip to main content

Aggregation - Reusability In COM-II

One way to reuse a component within another component is using containment. Another way to do so is using aggregation. In containment the outer component makes use of the inner component, but does not directly expose the inner component to its clients. The outer component can choose to use the inner component to implement some of its functionality. If the outer component decides to implement one or more of the interfaces of the inner component it must provide a wrapper for every method in the exposed interfaces. The method wrappers can simply call the inner component’s methods directly, or they can perform additional tasks such as validating parameters, recording usage, or otherwise extending the inner component’s interface.

Sometimes, it’s more convenient to expose the interfaces of the inner object directly, without writing a collection of wrappers. This is aggregation. A client creates the outer object, and when it asks QueryInterface( ) for an interface supported by the inner object, it gets passed a pointer to the inner object’s interface. When we use containment or aggregation the client does not know that there are actually two separate components acting in unison. We may as well have more than two objects. The outer component can aggregate several different components. Similarly, the inner component can aggregate another component into itself.

Aggregation has one distinct advantage over containment: it requires much less knowledge of the inner component on the part of the person implementing the outer component. Also, bypassing the wrapper layer makes the component that bit more efficient. The disadvantages of aggregation are that the inner (aggregated) object must be specially written so that it can be aggregated, and that the inner object can only be implemented in a DLL that is loaded directly into the same apartment as the outer object. There is no cross-apartment, cross-process or cross-machine aggregation.

Let us now understand aggregation with the help of some concrete example. Suppose we have a component that knows how to swap mouse buttons and set the time interval between two clicks of the mouse button. We can develop another component which uses this existing component and additionally also manage mouse trails. The existing component becomes the inner component. Given below is the procedure for building these components and a client that accesses the functionality of both the components.

Creating The Inner Component

The process of creating components in ATL consists of three steps:

(a) Creating Module:

To create a module the Developer Studio provides an ATL COM AppWizard. Carry out the following steps:

  • Select ‘New’ from the ‘File’ menu.

  • Select ‘ATL COM AppWizard’ as the project. Type ‘Inner’ as the project name and click ‘OK’ to continue.

  • Select type of module as ‘Dynamic Link Library’, click ‘Finish’.

(b) Adding Component To The Module

To add component to the module we can use ‘ATL Object Wizard’. Carry out the following steps for adding a component using this wizard:

  • Select ‘Insert | New ATL Object’ menu item. This would display the ‘ATL Object Wizard’

  • Select ‘Simple Object’ from the various object categories and click on ‘Next’.

  • A ‘ATL Object Wizard Properties’ dialog is displayed.

  • Enter the ‘Short Name’ as ‘Mouse’. As soon as you do this all other edit controls would be filled automatically. Click on OK.

(c) Adding Methods To The Component

The component that has been added does not contain any functionality. To provide functionality we should add two methods named SwapButton( ) and SetDoubleclicktime( ) as indicated below.

  • Switch to class view tab. Select the interface ‘IMouse’ and click the right mouse button. From the menu that pops up, select ‘Add Method’.

  • In the ‘Add Method to Interface’ dialog specify the method name as ‘SwapButton’ and leave the parameters edit control empty.

  • Click on ‘OK’.

  • Adding the SwapButton( ) method creates a function definition in ‘Mouse.cpp’ as shown below:

STDMETHODIMP CMouse::SwapButton( )
{

// TODO: Add your implementation code here

return S_OK ;

}

  • Add the following code to the SwapButton( ) method by double clicking on this method from the Class view tab:

mousestatus = !mousestatus ;

SystemParametersInfo ( SPI_SETMOUSEBUTTONSWAP, mousestatus,NULL,NULL ) ;

  • From the class view tab add a private variable mousestatus of type BOOL to the CMouse class. Initialize this variable to FALSE in the constructor of CMouse class.

  • On similar lines add another method called SetDoubleclicktime( ). Its code is given below:

STDMETHODIMP CMouse::SetDoubleclicktime ( long time )

{

AFX_MANAGE_STATE(AfxGetStaticModuleState( ) )

// TODO: Add your implementation code here

SystemParametersInfo ( SPI_SETDOUBLECLICKTIME, time, NULL, NULL ) ;

return S_OK ;

}

Creating The Outer Component

The procedure for creating the outer component is same as that for creating the inner component. First create a module called ‘Outer’. Insert a component called ‘MouseTrails’ in it and then add a method called SetMouseTrails( ) to it.

This methods is shown below:

STDMETHODIMP CMouseTrails::SetMouseTrails ( int trails )

{

AFX_MANAGE_STATE ( AfxGetStaticModuleState( ) )

// TODO: Add your implementation code here

SystemParametersInfo ( SPI_SETMOUSETRAILS, trails, NULL, NULL ) ;

return S_OK ;

}

  • Include the file ‘inner.h’ in ‘mousetrail.h’ as shown below:

# include "..\inner\inner.h"

  • Add a private member variable called m_agg in the outer class from the class view tab:

CComPtr m_agg ;

This variable would be used to hold the IUnknown pointer that we are going to get back from the aggregate when we create it.

  • Add the following macro to the file ‘MouseTrails.h’

DECLARE_GET_CONTROLLING_UNKNOWN( )

  • Add the following function to the ‘MouseTrails.h’ file.

HRESULT FinalConstruct( )

{

HRESULT hr ;

CLSID clsid ;

CoInitialize ( NULL ) ;

hr = CLSIDFromProgID ( OLESTR ( "inner.Mouse" ), &clsid )

if ( FAILED ( hr ) )

::MessageBox ( 0, "", "CLSIDFromProgID failed", 0 ) ;

hr = CoCreateInstance ( clsid, GetControllingUnknown( ), CLSCTX_INPROC, IID_IUnknown, ( void ** ) &m_agg ) ;

if ( FAILED ( hr ) )

::MessageBox ( 0, "", "CoCreate failed",0) ;

return S_OK ;

}

This function creates an instance of the inner object when the outer object first gets initialized. Here we are using CoCreateInstance( ) to create the instance of the CMouse coclass. The second parameter to CoCreateInstance( ) is the outer unknown of CMouseTrails object. We do not know that the current instance of the CMouse coclass is the controlling object, or whether it too has been aggregated. Hence we are using GetControllingUnknown( ) which will either return the controlling outer IUnknown if it is aggregated, or its own IUnknown if it is not. The fourth parameter is the interface we are asking for, which must be IUnknown when we are creating an object as part of an aggregate.

  • Add the following function to the ‘MouseTrails.h’ file.

HRESULT FinalRelease( )
{
inner->Release( ) ;
}

  • Add a COM_INTERFACE_ENTRY_AGGREGATE( ) macro for the IMouse interface to the com map.

COM_INTERFACE_ENTRY_AGGREGATE( IID_IMouse, m_agg.p )

This macro takes an interface identifier for the interface that we are exposing and an IUnknown * for the inner object that implements the interface. The macro itself hides the details that handle the implementation of QueryInterface( ).

  • Update the file outer.idl file as shown below:

library OUTERLib
{
importlib("stdole32.tlb");
importlib("stdole2.tlb");
importlib("..\inner\inner.tlb");
[
uuid(CB45BCB0-8A4A-11BD-A251-E0EAC39E8615),
helpstring("MouseTrails Class")
]
coclass MouseTrails
{
[default] interface IMouseTrails;
interface IMouse ;
};
};

Creating A Client

The steps to create a COM Client using MFC are as under:

  • Create a dialog-based project using AppWizard (EXE).

  • Add two edit boxes and three buttons to the dialog as shown in Figure 1.

Figure 1.

  • Add two variables m_trails and m_dctime of type integer for the two edit controls in the dialog box.

  • Import the Type Libraries into the client. This is done by adding the following statements to the file ‘StdAfx.h’.

# import "..\Outer\Outer.tlb"
using namespace OUTERLib ;
# import "..\Inner\Inner.tlb"
using namespace INNERLib ;

  • Add the regulation code in the OnInitDialog( ) function as shown below:

BOOL CContClientDlg::OnInitDialog( )
{

// AppWizard generated code
// TODO: Add extra initialization here
CLSID c ;
HRESULT h ;
CoInitialize ( NULL ) ;
h = CLSIDFromProgID ( OLESTR ( "Outer.MouseTrails" ), &c ) ;
if ( FAILED ( h ) )
MessageBox ( "CLSIDFromProgID Failed" ) ;
h = CoCreateInstance ( c, NULL, CLSCTX_ALL, __uuidof ( IMouseTrails ), ( void ** ) &outer ) ;

if ( FAILED ( h ) )
MessageBox ( " Not able to create instance of an object ") ;

hr = outer -> QueryInterface ( __uuidof ( IMouse ), (void **) &inner );
if ( FAILED ( hr ) )

MessageBox ( "Unable to query the inner interface" ) ;

inner -> SwapButtons = FALSE ;

return TRUE; // return TRUE unless you set the focus to a control

}

  • From the class view tab add private variables outer, inner of types IMouseTrails * and IMouse * respectively to the CClientDlg class.

  • Use the COM Object. Now that we have obtained the interface pointer, the client application can call the methods of the COM server object as shown below:

void CClientDlg::OnSwap( )
{
// TODO: Add your control notification handler code he
inner -> SwapButton( ) ;
}

void CClientDlg::OnMouseTrails( )
{
// TODO: Add your control notification handler code here
UpdateData ( TRUE ) ;
inner -> SetMouseTrails ( m_trails ) ;
}

void CClientDlg::OnDoubleClick( )
{
// TODO: Add your control notification handler code here
UpdateData ( TRUE ) ;
outer -> Setdoubleclicktime ( m_dctime ) ;
}

  • Uninitialize the COM library by the function CoUninitialize( ) at the end of InitInstance( ) function as shown below:

BOOL CContClientApp::InitInstance( )
{
// wizard generate code
CoUninitialize( ) ;
return FALSE;
}
Labels:

Comments

Post a Comment

Popular posts from this blog

MFC - Microsoft Foundation Classes Design Patterns

1 Introduction This paper describes the use of object-oriented software design patterns, as presented in Design Patterns: Elements of Reusable Object-Oriented Software by Gamma et al., within the Microsoft Foundation Class Library (MFC). MFC is used for implementing applications for Microsoft Windows operating systems. Because of the size of the MFC library, a complete analysis would have been beyond the scope of this assignment. Instead, we identified various possible locations for design patterns, using the class hierachy diagram of MFC, and studied the source carefully at these locations. When we did not find a pattern where we expected one, we have documented it anyway, with examples of how the particular problem could have been solved differently, perhaps more elegantly, using design patterns. We have included a brief introduction to MFC in Section 2 , as background information. The analysis has been split into three parts, with one section for each major design pattern ca...
1 comment
Read more

Explain Polymorphism and Flavors of Polymorphism...

Polymorphism is the ability of different objects to react in an individual manner to the same message. This notion was imported from natural languages. For example, the verb "to close" means different things when applied to different objects. Closing a door, closing a bank account, or closing a program's window are all different actions; their exact meaning is determined by the object on which the action is performed. Most object-oriented languages implement polymorphism only in the form of virtual functions. But C++ has two more mechanisms of static (meaning: compile-time) polymorphism: Operator overloading. Applying the += operator to integers or string objects, for example, is interpreted by each of these objects in an individual manner. Obviously, the underlying implementation of += differs in every type. Yet, intuitively, we can predict what results are. Templates. A vector of integers, for example, reacts differently from a vector of string objects when it receives ...
17 comments
Read more

• Why might you need exception handling be used in the constructor when memory allocation is involved?

Your first reaction should be: "Never use memory allocation in the constructor." Create a separate initialization function to do the job. You cannot return from the constructor and this is the reason you may have to use exception handling mechanism to process the memory allocation errors. You should clean up whatever objects and memory allocations you have made prior to throwing the exception, but throwing an exception from constructor may be tricky, because memory has already been allocated and there is no simple way to clean up the memory within the constructor.
Post a Comment
Read more