Liskov's Substitution Principle(LSP) All the time we design a program module and we create some class hierarchies. Then we extend some classes creating some derived classes. We must make sure that the new derived classes just extend without replacing the functionality of old classes. Otherwise, the new classes can produce undesired effects when they are used in existing program modules. Liskov's Substitution Principle states that if a program module is using a Base class, then the reference to the Base class can be replaced with a Derived class without affecting the functionality of the program module.

例子:

Below is the classic example for which the Liskov's Substitution Principle is violated. In the example, 2 classes are used: Rectangle and Square. Let's assume that the Rectangle object is used somewhere in the application. We extend the application and add the Square class. The square class is returned by a factory pattern, based on some conditions and we don't know the exact what type of object will be returned. But we know it's a Rectangle. We get the rectangle object, set the width to 5 and height to 10 and get the area. For a rectangle with width 5 and height 10, the area should be 50. Instead, the result will be 100

    // Violation of Likov's Substitution Principle
class Rectangle {
    protected int m_width;
    protected int m_height;

    public void setWidth(int width) {
        m_width = width;
    }

    public void setHeight(int height) {
        m_height = height;
    }

    public int getWidth() {
        return m_width;
    }

    public int getHeight() {
        return m_height;
    }

    public int getArea() {
        return m_width * m_height;
    }
}

class Square extends Rectangle {
    public void setWidth(int width) {
        m_width = width;
        m_height = width;
    }

    public void setHeight(int height) {
        m_width = height;
        m_height = height;
    }

}

class LspTest {
    private static Rectangle getNewRectangle() {
        // it can be an object returned by some factory ...
        return new Square();
    }

    public static void main(String args[]) {
        Rectangle r = LspTest.getNewRectangle();

        r.setWidth(5);
        r.setHeight(10);
        // user knows that r it's a rectangle.
        // It assumes that he's able to set the width and height as for the base
        // class

        System.out.println(r.getArea());
        // now he's surprised to see that the area is 100 instead of 50.
    }
}

结论: 这个原则只是开闭原则的延伸 意味着我们必须确保新的派生类正在扩展 基类而不改变它们的行为。

参见:开闭原则

对于更好的结构，还有一些类似的概念:约定优于配置

大局:

利斯科夫替换原理是什么?它是关于什么是(什么不是)给定类型的子类型。 为什么它如此重要?因为子类型和子类之间是有区别的。

例子

与其他答案不同的是，我不会从违反Liskov替换原则(LSP)开始，而是从符合LSP开始。我使用Java，但在每种OOP语言中几乎都是一样的。

圆和彩色圆

几何例子在这里似乎很流行。

class Circle {
    private int radius;

    public Circle(int radius) {
        if (radius < 0) {
            throw new RuntimeException("Radius should be >= 0");
        }
        this.radius = radius;
    }

    public int getRadius() {
        return this.radius;
    }
}

半径不允许为负。这是一个子类:

class ColoredCircle extends Circle {
    private Color color; // defined elsewhere

    public ColoredCircle(int radius, Color color) {
        super(radius);
        this.color = color;
    }

    public Color getColor() {
        return this.color;
    }
}

从LSP来看，这个子类是Circle的子类型。

LSP的状态为:

如果对于类型S的每个对象o1，都有一个类型T的对象o2，使得对于所有用T定义的程序P，当o1取代o2时，P的行为不变，那么S是T的子类型(Barbara Liskov，“数据抽象和层次结构”，SIGPLAN通知，23,5(1988年5月))。

这里，对于每个ColoredCircle实例o1，考虑Circle实例具有相同的半径o2。对于每个使用Circle对象的程序，如果您将o2替换为o1，则任何使用Circle的程序的行为在替换之后都将保持不变。(注意，这只是理论上的:使用ColoredCircle实例会比使用Circle实例更快地耗尽内存，但这与本文无关。)

我们如何根据o1求出o2 ?我们只是去掉color属性，保留radius属性。我称这个变换为o1 - >o2是CircleColor空间在Circle空间上的投影。

反例

让我们再创建一个例子来说明LSP的违反。

圆形和方形

想象一下前面Circle类的子类:

class Square extends Circle {
    private int sideSize;

    public Square(int sideSize) {
        super(0);
        this.sideSize = sideSize;
    }

    @Override
    public int getRadius() {
        return -1; // I'm a square, I don't care
    }

    public int getSideSize() {
        return this.sideSize;
    }
}

LSP违反

现在，看看这个程序:

public class Liskov {
    public static void program(Circle c) {
        System.out.println("The radius is "+c.getRadius());
    }

我们用一个Circle对象和一个Square对象测试程序。

    public static void main(String [] args){
        Liskov.program(new Circle(2)); // prints "The radius is 2"
        Liskov.program(new Square(2)); // prints "The radius is -1"
    }
}

发生了什么事?直观地说，虽然Square是Circle的一个子类，但Square不是Circle的子类型，因为没有一个常规的Circle实例的半径是-1。

形式上，这违反了利斯科夫替换原则。

我们有一个用Circle定义的程序，在这个程序中没有Circle对象可以替换新的Square(2)(顺便说一下，也没有任何Square实例)，并且保持行为不变:记住，任何Circle的半径总是正的。

子类和子类型

现在我们知道为什么子类并不总是子类型。当子类不是子类型时，即存在LSP违反时，某些程序(至少有一个)的行为并不总是预期的行为。这是非常令人沮丧的，通常被解释为一个错误。

在理想的情况下，编译器或解释器将能够检查给定的子类是否是真正的子类型，但我们并不是在理想的情况下。

静态类型

如果存在一些静态类型，则在编译时被父类签名绑定。Square.getRadius()不能返回String或List。

如果没有静态类型，如果一个参数的类型是错误的(除非类型是弱的)或参数的数量不一致(除非语言是非常允许的)，您将在运行时得到一个错误。

关于静态类型的注意:有返回类型的协方差(S的方法可以返回T的相同方法的返回类型的子类)和参数类型的逆变性(S的方法可以接受T的相同方法的相同参数的超类)的机制，这是下面解释的先决条件和后置条件的具体情况。

合同设计

有更多的。有些语言(我想到了Eiffel)提供了一种机制来强制执行LSP。

先不说确定初始对象o1的投影o2，如果用o1代替o2 if，对于任何参数x和任何方法f，我们可以期望任何程序都有相同的行为:

如果o2.f(x)是一个有效调用，那么o1.f(x)也应该是一个有效调用(1)。 o1.f(x)的结果(返回值，在控制台上显示等)应该等于o2.f(x)的结果，或者至少同样有效(2)。 o1.f(x)应该让o1处于内部状态，o2.f(x)应该让o2处于内部状态，这样下一次函数调用将确保(1)，(2)和(3)仍然有效(3)。

(注意，如果函数f是纯函数，则(3)是免费给出的。这就是为什么我们喜欢使用不可变对象。)

这些条件是关于类的语义(期望什么)，而不仅仅是类的语法。而且，这些条件非常强。但是它们可以用契约式编程设计中的断言来近似。这些断言是确保支持类型语义的一种方法。破坏契约会导致运行时错误。

前提条件定义了什么是有效的调用。当子类化一个类时，前提条件只能被削弱(S.f接受的比T.f多)(a)。 后置条件定义了什么是有效结果。当子类化一个类时，后置条件只能被加强(S.f比T.f提供更多)(b)。 不变量定义了什么是有效的内部状态。当子类化一个类时，不变量必须保持不变(c)。

我们可以大致看到，(a)保证了(1)，(b)保证了(2)，但是(c)比(3)弱。此外，断言有时难以表达。

假设一个类Counter有一个唯一的方法Counter. Counter()，该方法返回下一个整数。怎么写后置条件呢?假设一个类Random有一个方法Random.高斯()，该方法返回一个介于0.0和1.0之间的浮点数。如何编写后置条件来检查分布是否为高斯分布?这也许是可能的，但成本太高，我们将依赖于测试而不是后置条件。

结论

不幸的是，子类并不总是子类型。这可能会导致意想不到的行为——bug。

面向对象语言提供了避免这种情况的机制。首先在语法层面。在语义层面上也是如此，这取决于编程语言:可以使用断言在程序文本中编码一部分语义。但是，由您来确保子类是子类型。

还记得你什么时候开始学习OOP吗?“如果关系是is - a，那么使用继承”。另一种方式也是如此:如果使用继承，请确保关系是is - a。

LSP在比断言更高的级别上定义了什么是子类型。断言是确保LSP得到维护的有价值的工具。

当一些代码认为它正在调用类型T的方法时，LSP是必要的，并且可能在不知情的情况下调用类型S的方法，其中S扩展了T(即S继承、派生于超类型T，或者是超类型T的子类型)。

例如，当一个函数的输入形参类型为T时，调用(即调用)的实参值类型为S。或者，当一个类型为T的标识符被赋值类型为S时，就会发生这种情况。

val id : T = new S() // id thinks it's a T, but is a S

LSP要求T类型方法(例如Rectangle)的期望(即不变量)，当调用S类型方法(例如Square)时不违反此期望。

val rect : Rectangle = new Square(5) // thinks it's a Rectangle, but is a Square
val rect2 : Rectangle = rect.setWidth(10) // height is 10, LSP violation

即使是具有不可变字段的类型仍然有不变量，例如，不可变的矩形设置器期望维度被独立修改，但不可变的正方形设置器违背了这一期望。

class Rectangle( val width : Int, val height : Int )
{
   def setWidth( w : Int ) = new Rectangle(w, height)
   def setHeight( h : Int ) = new Rectangle(width, h)
}

class Square( val side : Int ) extends Rectangle(side, side)
{
   override def setWidth( s : Int ) = new Square(s)
   override def setHeight( s : Int ) = new Square(s)
}

LSP要求子类型S的每个方法必须有逆变的输入参数和协变的输出。

逆变是指方差与继承方向相反，即子类型S的每个方法的每个输入参数的Si类型必须与超类型T的相应方法的相应输入参数的Ti类型相同或为超类型。

协方差是指子类型S的每个方法的输出的方差在继承的同一方向，即类型So，必须是超类型T的相应方法的相应输出的相同或类型To的子类型。

这是因为如果调用者认为它有一个类型T，认为它正在调用一个类型T的方法，那么它就会提供类型Ti的参数，并将输出分配给类型to。当它实际调用S的对应方法时，每个Ti输入参数被赋值给Si输入参数，So输出被赋值给类型to。因此，如果Si与Ti的w.r.t.不是逆变的，那么就可以将Si的子类型xi赋给Ti，而它不是Si的子类型。

此外，对于在类型多态性参数(即泛型)上具有定义-站点方差注释的语言(例如Scala或Ceylon)，类型T的每个类型参数的方差注释的共方向或反方向必须分别与具有类型参数类型的每个输入参数或输出(T的每个方法)的方向相反或相同。

此外，对于每个具有函数类型的输入参数或输出，所需的方差方向是相反的。该规则是递归应用的。

---

子类型适用于可以枚举不变量的地方。

关于如何对不变量建模，以便由编译器强制执行，有很多正在进行的研究。

Typestate (see page 3) declares and enforces state invariants orthogonal to type. Alternatively, invariants can be enforced by converting assertions to types. For example, to assert that a file is open before closing it, then File.open() could return an OpenFile type, which contains a close() method that is not available in File. A tic-tac-toe API can be another example of employing typing to enforce invariants at compile-time. The type system may even be Turing-complete, e.g. Scala. Dependently-typed languages and theorem provers formalize the models of higher-order typing.

Because of the need for semantics to abstract over extension, I expect that employing typing to model invariants, i.e. unified higher-order denotational semantics, is superior to the Typestate. ‘Extension’ means the unbounded, permuted composition of uncoordinated, modular development. Because it seems to me to be the antithesis of unification and thus degrees-of-freedom, to have two mutually-dependent models (e.g. types and Typestate) for expressing the shared semantics, which can't be unified with each other for extensible composition. For example, Expression Problem-like extension was unified in the subtyping, function overloading, and parametric typing domains.

我的理论立场是，对于知识的存在(见章节“集中化是盲目的和不合适的”)，永远不会有一个通用模型可以在图灵完备的计算机语言中强制100%覆盖所有可能的不变量。要让知识存在，就必须存在许多意想不到的可能性，即无序和熵必须总是在增加。这是熵力。证明一个潜在扩展的所有可能的计算，就是计算一个先验的所有可能的扩展。

This is why the Halting Theorem exists, i.e. it is undecidable whether every possible program in a Turing-complete programming language terminates. It can be proven that some specific program terminates (one which all possibilities have been defined and computed). But it is impossible to prove that all possible extension of that program terminates, unless the possibilities for extension of that program is not Turing complete (e.g. via dependent-typing). Since the fundamental requirement for Turing-completeness is unbounded recursion, it is intuitive to understand how Gödel's incompleteness theorems and Russell's paradox apply to extension.

对这些定理的解释将它们纳入对熵力的广义概念理解中:

Gödel's incompleteness theorems: any formal theory, in which all arithmetic truths can be proved, is inconsistent. Russell's paradox: every membership rule for a set that can contain a set, either enumerates the specific type of each member or contains itself. Thus sets either cannot be extended or they are unbounded recursion. For example, the set of everything that is not a teapot, includes itself, which includes itself, which includes itself, etc…. Thus a rule is inconsistent if it (may contain a set and) does not enumerate the specific types (i.e. allows all unspecified types) and does not allow unbounded extension. This is the set of sets that are not members of themselves. This inability to be both consistent and completely enumerated over all possible extension, is Gödel's incompleteness theorems. Liskov Substition Principle: generally it is an undecidable problem whether any set is the subset of another, i.e. inheritance is generally undecidable. Linsky Referencing: it is undecidable what the computation of something is, when it is described or perceived, i.e. perception (reality) has no absolute point of reference. Coase's theorem: there is no external reference point, thus any barrier to unbounded external possibilities will fail. Second law of thermodynamics: the entire universe (a closed system, i.e. everything) trends to maximum disorder, i.e. maximum independent possibilities.

这里有一个清单来确定你是否违反了利斯科夫法则。

如果你违反了以下项目之一->，你违反了里斯科夫。 如果你不违反任何->不能得出任何结论。

检查表:

No new exceptions should be thrown in derived class: If your base class threw ArgumentNullException then your sub classes were only allowed to throw exceptions of type ArgumentNullException or any exceptions derived from ArgumentNullException. Throwing IndexOutOfRangeException is a violation of Liskov. Pre-conditions cannot be strengthened: Assume your base class works with a member int. Now your sub-type requires that int to be positive. This is strengthened pre-conditions, and now any code that worked perfectly fine before with negative ints is broken. Post-conditions cannot be weakened: Assume your base class required all connections to the database should be closed before the method returned. In your sub-class you overrode that method and left the connection open for further reuse. You have weakened the post-conditions of that method. Invariants must be preserved: The most difficult and painful constraint to fulfill. Invariants are sometimes hidden in the base class and the only way to reveal them is to read the code of the base class. Basically you have to be sure when you override a method anything unchangeable must remain unchanged after your overridden method is executed. The best thing I can think of is to enforce these invariant constraints in the base class but that would not be easy. History Constraint: When overriding a method you are not allowed to modify an unmodifiable property in the base class. Take a look at these code and you can see Name is defined to be unmodifiable (private set) but SubType introduces new method that allows modifying it (through reflection): public class SuperType { public string Name { get; private set; } public SuperType(string name, int age) { Name = name; Age = age; } } public class SubType : SuperType { public void ChangeName(string newName) { var propertyType = base.GetType().GetProperty("Name").SetValue(this, newName); } }

还有2项:方法参数的逆变性和返回类型的协方差。但这在c#中是不可能的(我是c#开发人员)，所以我不关心它们。

关于LSP的一个很好的例子(在我最近听到的播客中，Bob叔叔给出了一个例子)是，有时候在自然语言中听起来正确的东西在代码中却不太适用。

在数学中，正方形是长方形。实际上，它是矩形的专门化。“is a”使您想用继承来建模。然而，如果在代码中你从Rectangle派生出Square，那么Square应该可以在任何你想要Rectangle的地方使用。这就导致了一些奇怪的行为。

假设你在你的Rectangle基类上有SetWidth和SetHeight方法;这似乎完全合乎逻辑。然而，如果你的矩形引用指向一个正方形，那么SetWidth和SetHeight没有意义，因为设置一个会改变另一个来匹配它。在这种情况下，Square未能通过矩形的利斯科夫替换测试，并且让Square继承Rectangle的抽象是一个糟糕的抽象。

你们都应该看看其他用励志海报解释的无价的坚实原则。

Liskov's Substitution Principle(LSP) All the time we design a program module and we create some class hierarchies. Then we extend some classes creating some derived classes. We must make sure that the new derived classes just extend without replacing the functionality of old classes. Otherwise, the new classes can produce undesired effects when they are used in existing program modules. Liskov's Substitution Principle states that if a program module is using a Base class, then the reference to the Base class can be replaced with a Derived class without affecting the functionality of the program module.

例子:

Below is the classic example for which the Liskov's Substitution Principle is violated. In the example, 2 classes are used: Rectangle and Square. Let's assume that the Rectangle object is used somewhere in the application. We extend the application and add the Square class. The square class is returned by a factory pattern, based on some conditions and we don't know the exact what type of object will be returned. But we know it's a Rectangle. We get the rectangle object, set the width to 5 and height to 10 and get the area. For a rectangle with width 5 and height 10, the area should be 50. Instead, the result will be 100

    // Violation of Likov's Substitution Principle
class Rectangle {
    protected int m_width;
    protected int m_height;

    public void setWidth(int width) {
        m_width = width;
    }

    public void setHeight(int height) {
        m_height = height;
    }

    public int getWidth() {
        return m_width;
    }

    public int getHeight() {
        return m_height;
    }

    public int getArea() {
        return m_width * m_height;
    }
}

class Square extends Rectangle {
    public void setWidth(int width) {
        m_width = width;
        m_height = width;
    }

    public void setHeight(int height) {
        m_width = height;
        m_height = height;
    }

}

class LspTest {
    private static Rectangle getNewRectangle() {
        // it can be an object returned by some factory ...
        return new Square();
    }

    public static void main(String args[]) {
        Rectangle r = LspTest.getNewRectangle();

        r.setWidth(5);
        r.setHeight(10);
        // user knows that r it's a rectangle.
        // It assumes that he's able to set the width and height as for the base
        // class

        System.out.println(r.getArea());
        // now he's surprised to see that the area is 100 instead of 50.
    }
}

结论: 这个原则只是开闭原则的延伸 意味着我们必须确保新的派生类正在扩展 基类而不改变它们的行为。

参见:开闭原则

对于更好的结构，还有一些类似的概念:约定优于配置