Object-Oriented Programming “The Rest of the Story”, CS 4450 – Chapter 16

Topics What is OOP? Objects Without Classes ◦ “Dynamic inheritance” in Javascript, Python Contract Programming Multiple Dispatch

What is Object Oriented Programming? Encapsulation Inheritance Subtype Polymorphism ◦ or “equivalent”

Encapsulation Separating interface from implementation Different mechanisms ◦ Classes & Interfaces ◦ Nested functions and closures

The Role of Classes Most OO languages have some kind of class construct Classes serve a variety of purposes, depending on the language: ◦ Group fields and methods together ◦ Are instantiable: the running program can create as many objects of a class as it needs ◦ Serve as the unit of inheritance: derived class inherits from base class or classes Chapter SixteenModern Programming Languages5

Classes More purposes: ◦ Serve as a type: objects (or references to them) can have a class or superclass name as their static type ◦ House static fields and methods: one per class, not one per instance  Serve as a labeled namespace (scope); control the visibility of contents outside the class definition Chapter SixteenModern Programming Languages6

Without Classes Imagine an OO language with no classes With classes, you create objects by instantiating a class Without classes, you could create an object from scratch by listing all its methods and fields on the spot Or, you could clone an existing prototype object and then modify parts of it Chapter SixteenModern Programming Languages7

Chapter SixteenModern Programming Languages8 x = new Stack(); x = { private Node top = null; public boolean hasMore() { return (top!=null); } public String remove() { Node n = top; top = n.getLink(); return n.getData(); } … } y = x.clone(); y.top = null; With classes: instantiation Without classes: raw object creation Without classes: prototype cloning

Prototypes A prototype is an object that is copied to make similar objects When making copies, a program can modify the values of fields, and can add or remove fields and methods (“dynamic OOP”) Prototype-based languages (Self, Javascript to some degree) use this concept instead of classes Chapter SixteenModern Programming Languages9

Without Classes Instantiation is only one use of classes Other things prototype-based languages must do without: ◦ Classes as types: most prototype-based languages are dynamically typed ◦ Inheritance: “static inheritance” requires classes  Prototype languages use “dynamic inheritance”  aka “delegation” Chapter SixteenModern Programming Languages10

Inheritance Inheritance is a way to defer attribute or method selection to another type ◦ a “superclass” This is static inheritance ◦ since relationships among types are defined in source code There is another way…

Delegation “ Runtime Inheritance” Object (not type)-based inheritance Instead of a superclass, an object has a delegate ◦ aka “prototype” (just another object) ◦ Delegates are set at runtime Any unresolved name lookup defers to the delegate object ◦ and so on up the delegate chain… ◦ A similar effect to subtype polymorphism, but without types! ◦ See prototype.ds (Javascript), prototype.py

About Dynamic Typing An object may or may not be able to respond to a particular request (“message”) — no compile-time check Total freedom: program can try using any method for any object Polymorphism is a “given” in dynamically-typed languages ◦ Polymorphism is only a conscious concern in statically- typed languages ◦ “Polymorphism is a way gaining some of the freedom of dynamic type checking without giving up the benefits of static checking.” (p. 128) Chapter SixteenModern Programming Languages13

Contract Programming A Perspective for Polymorphic Interfaces Methods are contracts with the user Users must meet pre-conditions of a method ◦ What the method expects from the client ◦ Index in a certain range, for example Method guarantees certain post-conditions ◦ but only if the pre-conditions were met 14CS Defensive Programming

Parties in Contracts Clients and Suppliers Clients must satisfy pre-conditions ◦ They may benefit from post-conditions ◦ But don’t have to Suppliers must satisfy post-conditions ◦ They may assume the pre-conditions ◦ But don’t have to This affects inheritance… 15CS Defensive Programming

Liskov Substitutability Principle All derived objects must be able to substitute for a base class object ◦ In contexts where public methods of the base class are used Derived classes must not change the “rules of the game” CS Defensive Programming16

Contracts and Inheritance Contracts are set by the base class ◦ Derived classes must obey the base contract ◦ Otherwise substitutability is compromised Clients program to the contract ◦ By using and understanding the base class interface and its conditions ◦ And by using base/interface pointers  and letting polymorphism do its work invisibly Example: contravar.cpp, contravar.d 17CS Defensive Programming

Contracts and Inheritance Summary “The problem for instances of B is how to be perfectly substitutable for instances of A. The only way to guarantee type safety and substitutability is to be equally or more liberal than A on inputs, and to be equally or more strict than A on outputs.” – Wikipedia “Require no more; Promise no less” 18CS Defensive Programming

Dynamic Dispatch The Mechanics of Subtype Polymorphism In Java/C++/C#, the static type of a reference may be a superclass or interface of the actual runtime class At runtime, the language system must find the right method for the actual class ◦ the dynamic type That’s dynamic dispatch: the hidden, implicit branch-on-class to implement method calls Optional in C++ (must use virtual keyword) Chapter SixteenModern Programming Languages19

Multiple Dispatch All of the examples we’ve seen of subtype polymorphism (or delegation) key off a single type (or object) to find the right method ◦ p.f(x,y) considers only the dynamic type of p  x and y are not considered at runtime Some languages (CLOS, Groovy, Perl 6) use all types involved to resolve the method lookup ◦ “runtime overloading” ◦ No one class/object “owns” the method

Double Dispatch Example VWX A ✔✔ B ✔ C ✔ Definitions for f(): What is the “most derived” function for the calls: c.f(x), c.f(w)?

An Ordering Perspective List parameter combinations in “odometer order” (most general to most specific): A,V * A,W A,X * B,V B,W B,X * C,V * C,W C,X Reverse, keeping only existing methods: C,V B,X A,X A,V To dispatch, test parameters in the order above for is-a, using RTTI See doubledisp.cpp

Switching the Parameter Order V A * V B V C * W A W B W C X A * X B * X C X B * X A * V C * V A * See doubledisp-B.cpp

Multiple Dispatch Any number of hierarchies/parameters may be used Again, applicable methods are considered in “most derived” order Supported natively by CLOS ◦ Common Lisp Object System ◦ “Generic Methods”

Multiple Dispatch Example See multimeth.lsp, multimeth.cpp, multimeth.d ZVWX A ✔✔ B ✔ C ✔ YVWX A ✔ B ✔ C ✔