如何向类中添加可选功能？

Question

I'm currently trying to learn the ins and outs of XNA/C# but I'm currently stuck on a problem: I want to add optional functionality to a class and I have no idea how to even start. For clarity, here is my current code:

public class Actor
{
    //Fields
    private Vector3 m_location;

    //Properties
    public Vector3 Location
    {
        get { return m_location; }
        set { m_location = value; }
    }

    //Constructors
    public Actor(float x, float y, float z)
    {
        Location = new Vector3(x, y, z);
    }
    public Actor(Vector3 location)
    {
        Location = location;
    }
}

public class Actor2D : Actor
{
    //Fields
    private Texture2D m_texture;

    //Properties
    public Texture2D Texture
    {
        get { return m_texture; }
        set { m_texture = value; }
    }

    //Constructors
    public Actor2D(float x, float y, float z) : base(x, y, z) {}
    public Actor2D(Vector3 vector) : base(vector) {}

    //Methods
    public void Draw(SpriteBatch spriteBatch)
    {
        spriteBatch.Draw(Texture, new Vector2(Location.X, Location.Y), Color.White);
    }
    public void Load(ContentManager content, string asset)
    {
        Texture = content.Load<Texture2D>(asset);
    }
}

The logic here is that all Actors have to have a location but they don't necessarily have to have a texture as an Actor could be a light source, trigger, etc etc. Actor2D is similar in this except that it's designated as an Actor that uses 2D texture. What I want to do is have the ability to add functionality to either of these classes on an as needed basis. Say I want the Actor2D to be animated or as mentioned earlier I want to add a trigger or light source to an untextured Actor... What are some ideas that would help me accomplish these goals?

Answer 1

您可以尝试查看装饰器模式 ，它允许您“粘贴”对象的行为。

Answer 2

If you really want object "morphing" as in objects changing their set of methods at runtime, you're looking at mixins . (tutorial here )
Fair warning: it is not for the faint hearted.

Now since at compile-time you know all the types that will have to be available, use a variety of interfaces:

• IHasTexture
• IDrawable : IHasTexture (this means and IDrawable automatically has to implement IHasTexture, since you can't draw otherwise)
• ILoadable
• ... as many as you want.

Then provide an implementation for each combination you need. Furthermore, I think it would be best to put all this in your business layer if you have any.

Hope this helps,

Bab.

EDIT: example.

public interface IMovable { void Move(); }

public class MoveMixin : IMovable
{
   public void Move() { Console.WriteLine("I am moving"); }
}

public class SomeDude { }

public static class TestClass
{
   public static Test()
   {
      var myMixinConfiguration = MixinConfiguration.BuildFromActive()
           .ForClass<SomeDude>().AddMixin<MoveMixin>()
           .BuildConfiguration();
      using(myMixinConfiguration.EnterScope())
      {//In this scope, SomeDude implements IMovable, but only if you get an instance through the ObjectFactory
         var myMovableDude = ObjectFactory.Create<SomeDude>(ParamList.Empty);
         ((IMovable)myMovableDude).Move(); //This line compiles and executes just fine, even though the class declaration does not specify anything about Move() !!
      }
      //The SomeDude class no longer implements IMovable
   }
}

Answer 3

Instead of making Actor2D (and its likes) derived from Actor . Just change those derived classes to classes of their own, and, have one class (- Actor ) for every "Actor" with all of those properties. (An Actor2D property...) Then use whichever you need at runtime.

They will not take up that much resources because they will be pointing to null 's when not used.

EDIT:

class Actor
{
    public Stuff2D stuff2d = new Stuff2D();
    public StuffLightPower lightPower = new StuffLightPower();
    //...
}

class Stuff2D
{
    public int height;
    public int weight;
    //...
}

class StuffLightPower
{
    public int lumen;
    public Color color;
    //...
}

And then use it like this:

Actor actor1 = new Actor();

//First it's going to be a 2D.
actor1.stuff2d.weight = 3;
//...

//Now it's changed to a light source.
actor1.stuff2d = null;
actor1.lightPower.color = Color.White;
//...

//And back again.
actor1.lightPower = null;
actor1.stuff2d = new Stuff2D();
//...

Answer 4

So, after a bunch of thinking and searching on Google/StackOverflow. I've finally found my answer. Rather than using a traditional OOP approach I learned that I should be using a more Component-Oriented Programming approach. I found a really great link at Component based game engine design.

Of particular interest is the 5-part series on T=Machine describing Entity Systems which seems to be an approach which does exactly what I was looking for. For those interested, the first post is: http://t-machine.org/index.php/2007/09/03/entity-systems-are-the-future-of-mmog-development-part-1/ .