Sebesta Chapter 12

12.1 Introduction

Categories of languages that support OOP:

1. OOP support is added to an existing language
C++ (also supports procedural and data-oriented programming)
Ada 95 (also supports procedural and data-oriented programming)
CLOS (also supports functional programming)
Scheme (also supports functional programming)


2. Support OOP, but have the same appearance and use the basic structure of earlier imperative languages

Eiffel (not based directly on any previous language)
Java (based on C++)
3. Pure OOP languages
Smalltalk
12.2 Object-Oriented Programming

 Paradigm Evolution

1. Procedural - 1950s-1970s (procedural abstraction)
2. Data-Oriented - early 1980s (data abstraction)
3. OOP - late 1980s (Inheritance and dynamic binding)
Origins of Inheritance
Observations of the mid-late 1980s :
Productivity increases can come from reuse
Unfortunately,
ADTs are difficult to reuse--never quite  right
All ADTs are independent and at the same  level
Inheritance solves both--reuse ADTs after minor changes and define classes in a hierarchy
OOP Definitions:
ADTs are called classes

Class instances are called objects

A class that inherits is a derived class or a subclass

The class from which another class inherits is a parent class or superclass

Subprograms that define operations on objects are called methods

The entire collection of methods of an object is called its message protocol or message interface

Messages have two parts--a method name and the destination object

In the simplest case, a class inherits all of the entities of its parent
 

Inheritance can be complicated by access  controls to encapsulated entities

A class can hide entities from its subclasses
A class can hide entities from its clients
A class can also hide entities for its clients while allowing its subclasses to see them
Besides inheriting methods as is, a class can  modify an inherited method
The new one overrides the inherited one
The method in the parent is overriden
There are two kinds of variables in a class:
1. Class variables - one/class
2. Instance variables - one/object
There are two kinds of methods in a class:
1. Class methods – accept messages to the
2. Instance methods – accept messages to objects
Single vs. Multiple Inheritance


One disadvantage of inheritance for reuse:

Creates interdependencies among classes that complicate maintenance
Polymorphism in OOPLs
A polymorphic variable can be defined in a class that is able to reference (or point to) objects of the class and objects of any of its descendants

When a class hierarchy includes classes that override methods and such methods are called
 through a polymorphic variable, the binding to the correct method MUST be dynamic

This polymorphism simplifies the addition of  new methods

A virtual method is one that does not include a definition (it only defines a protocol)

A virtual class is one that includes at least one virtual method

A virtual class cannot be instantiated

Design Issues for OOPLs
 1. The Exclusivity of Objects

    a. Everything is an object
        Advantage - elegance and purity
        Disadvantage - slow operations on simple   objects (e.g., float)
    b. Add objects to a complete typing system
         Advantage - fast operations on simple objects
         Disadvantage - results in a confusing type     system (two kinds of entities)
    c. Include an imperative-style typing system for
         primitives but make everything else objects
         Advantage - fast operations on simple objects and a relatively small typing system
         Disadvantage - still some confusion because of the two type systems
 

 2. Are Subclasses Subtypes?

    - Does an is-a relationship hold between a parent  class object and an object of the subclass?
3. Implementation and Interface Inheritance

  - If only the interface of the parent class is visible  to the subclass, it is interface inheritance
 
     Disadvantage - can result in inefficiencies

  - If both the interface and the implementation of the parent class is visible to the subclass, it is  implementation inheritance
Disadvantage - changes to the parent class  require recompilation of   subclasses, and sometimes even  modification of subclasses

4. Type Checking and Polymorphism

   - Polymorphism may require dynamic type
     checking of parameters and the return value
       - Dynamic type checking is costly and delays
          error detection
   - If overriding methods are restricted to having the
      same parameter types and return type, the
      checking can be static
5. Single and Multiple Inheritance

   - Disadvantages of multiple inheritance:
      - Language and implementation complexity (in
         part due to name collisions)

      - Potential inefficiency - dynamic binding costs
        more with multiple inheritance (but not much)

   - Advantage:
      - Sometimes it is extremely convenient and
         valuable
 

 6. Allocation and Deallocation of Objects

   - From where are objects allocated?
     - If they all live in the heap, references to them
        are uniform

        - Simplifies assignment - dereferencing can be
           implicit

   - Is deallocation explicit or implicit?

7. Dynamic and Static Binding

   - Should ALL binding of messages to methods be
      dynamic?

      - If none are, you lose the advantages of
         dynamic binding

      - If all are, it is inefficient
12.4 Overview of Smalltalk

- Smalltalk is a pure OOP language
  - Everything is an object
  - All computation is through objects sending
     messages to objects
  - It adopts none of the appearance of imperative
     languages

- The Smalltalk Environment
   - The first complete GUI system
   - A complete system for software development
   - All of the system source code is available to
      the user, who can modify it if he/she wants
 

12.5 Introduction to Smalltalk

- Expressions
  - Four kinds:
      1. Literals (numbers, strings, and keywords)
 
      2. Variable names (all variables are references)

      3. Message expressions (see below)

      4. Block expressions (see below)
- Message expressions

   - Two parts: the receiver object and the message
      itself
   - The message part specifies the method and
      possibly some parameters
   - Replies to messages are objects

 - Messages can be of three forms:
   1. Unary (no parameters)
       e.g., myAngle sin
       (sends a message to the sin method of the
        myAngle object)

    2. Binary (one parameter, an object)
       e.g., 12 + 17
        (sends the message “+ 17” to the object 12;
          the object parameter is “17” and the method
          is “+”)

    3. Keyword (use keywords to organize the
         parameters)
        e.g., myArray at: 1 put: 5
         (sends the objects “1” and “5” to the at:put:
          method of the object myArray)

 - Multiple messages to the same object can be
    strung together, separated by semicolons
- Methods

   - General form:
     message_pattern [| temps |] statements

   - A message pattern is like the formal parameters
      of a subprogram
      - For a unary message, it is just the name
      - For others, it lists keywords and formal names
   - temps are just names--Smalltalk is typeless!
 

 - Assignments

  - Simplest Form:
     name1 <- name2

  - It is simply a pointer assignment

  - RHS can be a message expression
    e.g., index <- index + 1
 

 - Blocks

   - A sequence of statements, separated by periods,
      delimited by brackets
      e.g.,
    [index <- index + 1. sum <- sum + index]
- Blocks (continued)

   - A block specifies something, but doesn’t do it
   - To request the execution of a block, send it the
      unary message, value
      e.g., […] value

   - If a block is assigned to a variable, it is evaluated
     by sending value to that variable
     e.g.,
      addIndex <- [sum <- sum + index]
     …
    addIndex value

   - Blocks can have parameters, as in
      [:x :y | statements]

   - If a block contains a relational expression, it
      returns a Boolean object, true or false
 

 - Iteration

   - The objects true and false have methods for
    building control constructs

   - The method WhileTrue: from Block is used for
     pretest logical loops. It is defined for all blocks
     that return Boolean objects.
- Iteration (continued)

e.g.,
  [count <= 20]
    whileTrue [sum <- sum + count.
               count <- count + 1]

 - timesRepeat: is defined for integers and can be
    used to build counting loops
   e.g.,
     xCube <- 1.
     3 timesRepeat: [xCube <- xCube * x]
 
 

 - Selection

 - The Boolean objects have the method
    ifTrue:ifFalse: , which can be used to build
    selection
    e.g.,
      total = 0
     ifTrue: […]
     ifFalse: […]

12.6 Smalltalk Example Programs

 - See Book, pp. 480-485

12.7 Large-Scale Features of Smalltalk

 - Type Checking and Polymorphism

   - All bindings of messages to methods is dynamic

      - The process is to search the object to which
         the message is sent for the method; if not
         found, search the superclass, etc.

   - Because all variables are typeless, methods are
      all polymorphic

  - Inheritance

    - All subclasses are subtypes (nothing can be
       hidden)

    - All inheritance is implementation inheritance

    - No multiple inheritance

    - Methods can be redefined, but the two are not
       related
12.9 Support for OOP in C++

 - General Characteristics:
 
   - Mixed typing system
   - Constructors and destructors
   - Elaborate access controls to class entities
 

 - Inheritance

   - A class need not be the subclass of any class

   - Access controls for members are
      1. Private (visible only in the class and friends)
           (disallows subclasses from being subtypes)
      2. Public (visible in subclasses and clients)
      3. Protected (visible in the class and in
          subclasses, but not clients)

       - In addition, the subclassing process can be
          declared with access controls (private or
          public), which define potential changes in
          access by subclasses

         a. Private derivation - inherited public and
               protected members are private in the
               subclasses
b. Public derivation public and protected
            members are also public and protected in
            subclasses

Example (book, p. 490)

class base_class {
  private:
    int a;
    float x;
  protected:
    int b;
    float y;
  public:
    int c;
    float z;
};

class subclass_1 : public base_class { … };

//  - In this one, b and y are protected and
//     c and z are public

class subclass_2 : private base_class { … };

// - In this one, b, y, c, and z are private,
//    and no derived class has access to any
//    member of base_class

- Reexportation

   A member that is not accessible in a subclass
    (because of private derivation) can be declared
    to be visible there using the scope resolution
    operator (::)

    e.g.,

     class subclass_3 : private base_class {
        base_class :: c;
    …
     }
 

  - One motivation for using private derivation:
     - A class provides members that must be visible,
       so they are defined to be public members;
       a derived class adds some new members, but
       does not want its clients to see the members
       of the parent class, even though they had to be
       public in the parent class definition
 

- Multiple inheritance is supported

      - If there are two inherited members with the
         same name, they can both be reference
         using the scope resolution operator
- Dynamic Binding

   - A method can be defined to be virtual, which
      means that they can be called through
      polymorphic variables and dynamically bound
      to messages

   - A pure virtual function has no definition at all

     - A class that has at least one pure virtual
        function is an abstract class
 

- Evaluation

   - C++ provides extensive access control (unlike
      Smalltalk)

   - C++ provides multiple inheritance

   - In C++, the programmer must decide at design
      time which methods will be statically bound
      and which must be dynamically bound
       - Static binding is faster!

    - Smalltalk type checking is dynamic (flexible,
       but somewhat unsafe)
    - Because of interpretation and dynamic binding,
       Smalltalk is ~10 times slower than C++
12.10 Support for OOP in Java

 - Because of its close relationship to C++, we focus
    on the differences from that language

 - General Characteristics

   - All data are objects except the primitive types

   - All primitive types have wrapper classes that
      store one data value

   - All objects are heap-dynamic, are referenced
      through reference variables, and most are
      allocated with new
 

- Inheritance

  - Single inheritance only, but there is an abstract
     class category that provides some of the
     benefits of multiple inheritance (interface)

     - An interface can include only method
        declarations and named constants

       e.g.,
         public class Clock extends Applet
                       implements Runnable

   - Methods can be final (cannot be overriden)
- Dynamic Binding

    - In Java, all messages are dynamically bound to
      methods, unless the method is final (means it
      cannot be overriden; therefore, dynamic binding
      serves no purpose)
 
 

 - Encapsulation

    - Two constructs, classes and packages

    - Packages provide a container for classes that
      are related (can be named or unamed)

    - Entities defined without a scope (access)
      modifier have package scope, which makes
      them visible throughout the package in which
      they are defined - they go in the unnamed
      package

      - Every class in a package is a friend to the
        package scope entities elsewhere in the
        package

        - So, package scope is an alternative to the
           friends of C++

12.11 Support for OOP in Ada 95

 - General Characteristics

   - OOP was one of the most important extensions
      to Ada 83

   - Encapsulation container is a package that
      defines a tagged type

   - A tagged type is one in which every object
      includes a tag to indicate during execution its
      type (the tags are internal)

   - Tagged types can be either private types or
      records

   - No constructors or destructors are implicitly
      called
 

- Inheritance

   - Subclasses can be derived from tagged types

   - New entities in a subclass are added in a record

- Example (of a tagged type)

Package PERSON_PKG is
  type PERSON is tagged private;
  procedure DISPLAY(P : in out PERSON);
  private
    type PERSON is tagged
      record
        NAME : STRING(1..30);
        ADDRESS : STRING(1..30);
        AGE : INTEGER;
      end record;
end PERSON_PKG;
 

with PERSON_PKG; use PERSON_PKG;
 package STUDENT_PKG is
   type STUDENT is new PERSON with
     record
       GRADE_POINT_AVERAGE : FLOAT;
       GRADE_LEVEL : INTEGER;
     end record;
   procedure DISPLAY (ST: in STUDENT);
 end STUDENT_PKG;
 

 - DISPLAY is being overriden from PERSON_PKG

- Inheritance (continued)

 - All subclasses are subtypes

 - Single inheritance only, except through generics
 
 

 - Dynamic Binding

  - Dynamic binding is done using polymorphic
     variables called classwide types

     e.g., for the tagged type PERSON, the classwide
          type is PERSON’class

  - Other bindings are static

  - Any method may be dynamically bound

12.12 Support for OOP in Eiffel

- General Characteristics

  - In the original version, all objects were heap
    allocated
    - Now, Eiffel has both those objects and “expanded
      objects” (stack allocated)

       - Variables can be references to objects or
          directly reference expanded objects

       - This provides the efficiency of direct addressing
          for primitives without requiring two typing
          systems

   - All objects get three operations, copy, clone,
      and equal

   - Methods are called routines

   - Instance variables are called attributes

   - The routines and attributes of a class are
      together called its features

   - Object creation is done with an operator (!!)

   - Constructors are defined in a creation clause,
      and are explicitly called in the statement in
      which an object is created

- Inheritance

    - The parent of a class is specified with the
        inherit clause
 

  - Access control
 
     - feature clauses specify access control to the
        entities defined in them

        - Without a modifier, the entities in a feature
           clause are visible to both subclasses and
           clients

        - With the name of the class as a modifier,
           entities are hidden from clients but are
           visible to subclasses

        - With the none modifier, entities are hidden
           from both clients and subclasses

   - Inherited features can be hidden from
      subclasses with undefine

   - Abstract classes can be defined by including the
      deferred modifier on the class definition
- Dynamic Binding

   - Nearly all message binding is dynamic

   - An overriding method must have parameters
      that are assignment compatible with those of
      the overriden method

   - All overriding features must be defined in a
      redefine clause

     - Access to overriden features is possible by
        putting their names in a rename clause
 

 - Evaluation

   - Similar to Java in that procedural programming
      is not supported and nearly all message
      binding is dynamic

    - Elegant and clean design of support for OOP

12.13 The Object Model of JavaScript

 - General Characteristics of JavaScript

    - Has little in common with Java

    - Dynamic typing

    - No classes or inheritance or polymorphism

    - Variables can reference objects or can directly
       access primitive values

 - JavaScript Objects

    - An object has a collection of properties, which
       are either data properties or method properties

    - Appear as hashes, both internally and externally
      - A list of property/value pairs

      - Properties can be added or deleted dynamically

      - A bare object can be created with new and a
         call to the constructor for Object

         var my_object = new Object();

      - References to properties are with dot notation

12.14 Implementing OO Constructs

 - Class instance records (CIRs) store the state of
    an object

 - If a class has a parent, the subclass instance
    variables are added to the parent CIR

 - Virtual Method Tables (VMTs) are used for
    dynamic binding