Using properties in managed C++

• Download demo application (beta 2) - 27 Kb

Introduction

With the introduction of managed extensions for C++ comes the concept of properties. A property is similar to a field, in that it allows you to read and write values to a class instance, but whereas a field is an actual data member that you are accessing directly, a property is implemented via get and set methods that allow you to fully control access to the value. This control can be extended to allow the property to be read-only, write-only, or read and write accessible, and allows you to perform processing such as validation or caching to simplify your code. A property does not need to be associated with an underlying data member of the class.

This article assumes you are familiar with declaring and using managed types and the .NET Garbage Collector.

C++ with managed extensions supports two types of properties: Scale and Indexed. Scalar Properties are analogous to member fields, and Indexed properties are analogous to member arrays.

Declaring a Scalar property

Property accessor functions use a special naming convention:

__property type get_PropertyName();           // declare a readable property

__property void set_PropertyName(type value); // declare a writeable property

type is the property's type, PropertyName is the name of the property, and __property is a new C++ keyword that declares the method as a property accessor. The property is accessed by referring to only the name of the property using the same syntax as you would if you were accessing a data field.

For example, suppose we have a managed class Item three properties: Cost, Size and Faulty. You wish to have Size be read-only, since it will never change, Cost be read/write accessible since it will change, and Faulty be write-accessible publicly, but not read-accessible from outside the class.

To control the accessibility of the properties you simply pick and choose which property methods (get/set) you implement. For the Size property we only want to implement a get_Size method. For Cost we want to implement get_Cost and set_Cost, and for Faulty we need set_Faulty:

__gc class Item
{
public:
    Item()
    {
        m_dCost = 0.0;
        m_blnFaulty = false;
    }
    
    // A write-only property called Faulty

    __property void set_Faulty(bool value)
    {
        m_blnFaulty = value;
    }

    // A read/write propert called Cost.

    __property decimal get_Cost()
    {
        return m_dmCost;
    }
    __property void set_Cost(decimal cost)
    {
        if (cost > 0)
            m_dmCost = cost;
        else
            m_dmCost = 0;
    }

    // A read only property called Size

    __property int get_Size()
    {
        return 100;
    }

private:
    decimal m_dmCost;
    bool    m_blnFaulty;
};

Notice that in the above class we have performed validation on the Cost property to ensure that illegal values aren't set, and we have implemented the Size property without associating a member variable. We could just as easily have implemented the Size property by having the class lookup a database or query a cache.

Note also the use of the new .NET type decimal.

Using Scalar Properties

To the consumer of the class all they see are properties that they can access like fields:

Item* item = new Item();

// Access the Read-only property as if it were a field

int Size = item->Size;

// Access the Write-only property as if it were a field

item->Faulty = true;

// The following line will produce a compiler error since we are 

// trying to access Faulty as a readable property, when it is write-only.

// if (item->Faulty) Console::WriteLine("Item is faulty");


// Access the Read/Write property as if it were a field

item->Cost = -4;

// show the value of Cost. It will be 0 due to range-checking

// performed in the set_ accessor.

Console::WriteLine("Item::Cost = {0}", item->Cost.ToString());

Note that the ToString() function in the above code simply converts the value types (int, double) into a string object containing a representation of the value. Value types in .NET can have member functions.

Declaring an Indexed property

Indexed properties are similar to scalar properties except that instead of accessing a single value as you would a field, you access one of many possible values as if the property were an array.

For instance, we will add a property to our Item class called Stock. This property is indexed as a 2-dimensional array by State and City (both declared as .NET String types):

// Indexed property - Get

__property int get_Stock(String* State, String* City)   
{
    // Return a value - either from an internal member array

    // or maybe a database query

   ...
}

// Indexed property - Set

__property void set_Stock(String* State, String* City, int value) 
{
   // Store the value in some meaningful way

   ...
}

Using an indexed property is the same as accessing an array member field of the class.

// Set item in Seattle, WA

item->Stock["Seattle"]["WA"] = 1;

// Get Stock at Toronto, ON

int nStock = item->Stock["Toronto"]["ON"];

Conclusion

Properties help promote encapsulation and can make you code neater and easier to to follow, but are only available in managed C++ programs. A few points about properties:

• You cannot take the address of a property
• The name of the Get and Set methods must start with get_ and set_ respectively, and be prefixed with the __property keyword.
• The names of the Get and Set methods for a given property must be the same apart from the get_ and set_ prefixes.
• Property methods can be virtual and pure virtual
• The Get and Set method may have different accessibility
• If either of the Get or Set methods are static, they both must be.
• You don't have to implement both the Get and Set methods (this controls read/write as stated above).
• The last variable type in the Set method's parameter list must match the return type of the Get method.

History

16 Oct 2001 - updated source files for VS.NET beta 2