Interface definition overview

It was stated earlier that an interface definition serves two purposes:

• defining functions that offer services to clients

• offering object instantiation functions

So, an interface definition class is a normal abstract class plus some implemented functions that use ECom to create object instances. The class can have any base class: for example,

• for CBase-based classes, you can implement the object construction in NewL() and destruction in the destructor

• for a RHandleBase-based class, the Open() and Close() functions would be the natural place to conceal the implementation instantiation and destruction

The class must always declare a private UID member. This is set on object instantiation and is used by ECom to identify the instance when object destruction occurs.

Example

The following shows an example segment of a interface definition class that derives from CActive (i.e. it is an active object).

class CExampleInterfaceDefinition : public CActive
    {
public:
    // Wraps ECom object instantitation
    static CExampleInterfaceDefinition* NewL();
    // Wraps ECom object destruction 
    virtual ~CExampleInterfaceDefinition();
    // Interface service 
    virtual void DoMethodL() = 0;
...
private:
    // Instance identifier key
    TUid iDtor_ID_Key;
    };

Notes

The class:

• Has a pure virtual method: this means that this class must be a base, although a NewL() usually implies that there is no derivation from this class.

• Does not need a constructor, private or otherwise, because for an abstract base there is no requirement for construction.

• Has no ConstructL() to accompany the NewL(), construction isn't going to happen in the standard two-phase pattern, but through ECom.