FCI-Helwan blog

Just another FCI-H weblog

OOP Design Concepts – Interfaces

What is an interface in OOP?
During the process of designing an application using OOP Paradigm, the application will be designed as a set of classes indicating how objects will be created each with specific status and behavior, as well as defining the ways of interaction between those objects. By this way, almost every class indicates a type(an abstracted data type in a more specific manner).
For example, a class named car indicates an object of type car that has a set of status variables and a set of methods that indicates how other objects can interact with it. Consider that we have a car that have the normal four wheels and another type that uses one wheel in the middle of the car to move!!.
The common behavior between the two types of cars is that they can move, either on 4 wheels or 1 wheel it does not make a difference, all what we care about is that they can move, so, simply we can consider the two cars as “Movable”. The word movable indicates the ability of the two objects to move which is an indication of the behavior of them without concerning other details like; the model of the car or even the way they can move. These are interfaces.
Interfaces indicate how certain types of objects have to behave. If we want to indicate clearly how a movable object must behave, we will create an interface with a method named ” move” and make every movable object provide how this method will be implemented.
In our example; we will create something like what follows:

public interface iMovable
{
public void move();
}

public class normalCar implements iMovable
{
//actual implementation of the move method declared in iMovable
public void move()
{
//the code required to make a normal car move
}
}

public class weirdCar implements iMovable
{
public void move()
{
//the weird code to make a weird car move
}
}

So, why do we need interfaces?
OOP tries to resemble how objects are defined in the real life, and interfaces are a very logical way of grouping objects in terms of behavior. Suppose that we are making a game that has the normal and the weird car,however, a creative team member came with the idea that we need to make the extra super weird car with no wheels and with the ability to turn to a frog to hide from enemies and many many extra features. If we are not using interfaces, the game logic will need to handle the motion of each car as if they are of different type in spite that logically they share the core common behavior.
Interfaces also enhance abstraction which is a core principal of OOP, design patterns like the factory pattern make a perfect use of interfaces through abstracting objects to know which implementation of a certain object type is provided. Interfaces make it very flexible to change implementations of services specially in multilayer applications, for example; in an application, we have a data base layer that have an object named UserDb which is responsible to get the data of the user from the database, and we have another one named UserTxtFile which retrieves the information from a plain text file.
Each of the user objects can be used by a higher layer to retrieve the user data without any modification to the code, inspect the following code:

public interface iUserInfo
{
string getUserName(int userId);
string getUserAge(string username);
}

public class UserDb implements iUserInfo
{
string getUserName(int userId)
{
//code to retrieve user info from the database
}
}

public class UserTxtFile implements iUserInfo
{
string getUserName(int userId)
{
//code to retrieve user info from the plain text file
}
}

public class UserManager()
{
private iUserInfo userInfo;

public string retrieveUserData(string userName)
{
return userInfo.getUserAge(userName);
}
}

Take a close look at the manager class and you will find that it has a reference variable of type iUserInfo, so, if this reference is provided by a factory method or by any other means, the code of the manager class will be the same regardless what implementation is provided. This approach maximizes flexibilty if some implementations need to be altered during the process of software development, as higher level layers will not be affected at all.
Another benefit of interfaces shows up during the development of large projects by large teams, at which the interface acts as a specification for developers with the set of methods they need to implement in classes of a certain type.

What interfaces are NOT?
– Unless the interface specifies a common behavior, do not create one. A common mistake is to use interfaces just to make some constants visible to objects in different layers.
– Interfaces are not a work around to multiple inhertience.

I hope i could make my points clear enough, and about the super car that turns to a frog ; i think it is a good idea. think about it 🙂 .

Advertisements

March 5, 2008 - Posted by | Concepts

5 Comments »

  1. very nice discussion.
    sorry i have a question that may be out of your topic
    i need to know what’s the difference between interface and abstract class and when to use each of them?
    thanks.

    Comment by soli | March 7, 2008 | Reply

  2. hey , its very nice work but i think i have kinda misunderstanding here . If interfaces are indicators to the behavior without concerning abt details how can it help in the example of the extra super wired car i mean i will have to implement the movement in each class thus if i have WiredCar class i will implement its way of movement ,
    Car1 which move normally i will implement its way of movement
    car2(assuming it share the same way of movement with car1 & have different behaviors than it)i will also have to implement its movement although its the same as car1
    how interfaces helped me here ???? by just saying
    that i have a method movement that is shared by the classes
    implements it ?????

    sorry 4 the long comment

    thnx in advance

    Comment by Samar | April 19, 2008 | Reply

  3. I’m not the author of this post, but I guess I can reply here..

    In that case, Samar, you’ll use abstract classes to implement the common behavior..

    That’s on the most abstract level, you’ll have your interface, below that in a bit less abstraction level you have your abstract classes which has the common behavior among your lowest level of abstraction (that’s WierdCar for eg)..

    Comment by sh@dy | April 19, 2008 | Reply

  4. thanks shady for ur reply, it is exactly as shady said, it is all about levels of abstraction, and another thing…suppose u have a player that has a car..the class of the player will have a member variable of type car (the interface type) and as long as the game continues he may buy a normal car or a weird one, at either cases, u know that when u want to move the player’s car u will find a method named move no matter what implementation class supplied for the car..got me??, u know the general type, and u need anything more….
    if we make another car it will extend the same hierarchy and providing the same behavior and the game logic will deal with it the same way as it deals with it as it deals with any other car…
    if it is still spaghetti feel free to ask as much as u want…..thanks

    Comment by Youssef Mamdouh | April 20, 2008 | Reply

  5. Very good article.
    There is one point that you made was obvious: alot of programmer just fancily use interface for multiple inheritance as a workaround. That causes a big frustrated code.
    Good job!

    Comment by Jinn | March 17, 2009 | Reply


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: