Copyright (C) 2018, H. Conrad Cunningham

Acknowledgements: I created these slides in Fall to accompany Chapter 3, Object-Based Paradigms, of the book Exploring Languages with Interpreters and Functional Programming.

Browser Advisory: The HTML version of this document may require use of a browser that supports the display of MathML. A good choice as of August 2018 is a recent version of Firefox from Mozilla.

Lecture Goals

• Examine general object model for object-oriented programming paradigm

• Examine related paradigms: object-based, class-based, and prototype-based

Assumptions

Key idea: Real world can be accurately described as collection of interacting objects

1. Easier to understand complex systems as interacting objects
2. Behaviors of real world objects stable over time
3. Kinds of real world objects stable over time
4. Changes tend to be localized to a few objects

Object-Oriented Approach

• Uses same basic entities (objects) throughout software development lifecycle

• Identifies basic objects during analysis

• Identifies lower-level objects during design, reusing existing object descriptions where appropriate

• Implements objects as software structures (classes)

• Maintains object behaviors

Object Model

1. Objects (abstract data structures)

2. Classes (abstract data types)

3. Inheritance (hierarchical relationships among abstract data types)

4. Subtype polymorphism

Dynamic binding? Consider implementation technique for polymorphism rather than characteristic

Model: Objects

Essential characteristics

• [a] state

• [b] operations

• [c] identity

Important but non-essential characteristics

• [d] encapsulation

• [e] independent lifecycle

First-Class Citizens

Objects usually first-class entities

• store in data structures

• pass to / return from procedures and functions

[a] State

Collection of information held by object:

• can change over time

• can change only as result of operation on object

• cannot change spontaneously

Attributes – components of state

[b] Operations

• Operation – procedure takes object state & arguments, changes state or returns values

  Object permits certain operations, not others

• Mutable object – operation can change stored state in-place – imperative language

• Immutable object – operation cannot change stored state in-place, returns modified object with new state

[c] Identity

Can distinguish between two distinct objects

• even with same state and operations

• perhaps in different physical locations (e.g. memory address)

[d] Encapsulation

Encapsulation – private features not directly accessible outside

E.g. Java private instance variables and method encapsulated

Object-based languages without true encapsulation

• Python – uses obfuscation, programming conventions

• Lua – uses programming conventions, advanced programming techniques

• Oberon – uses modules for encapsulation, extended record types for other OO features

[e] Independent Lifecycle

Independent lifecycle:

Object exists independently from program unit that created it (independent lifetime)

• Usually means dynamic allocation on heap

• But some languages allow object on stack – e.g., C++

Classroom Sim Objects

• Each StudentDesk distinct from others – unique identity

• State – location, orientation, person using, items in basket, items on top

• State-changing operations – mutator, setter, command

  “move desk”, “seat student”, “remove from basket”

• State-observing operations (accessor, getter, observer, query)

  “is occupied”, “report items on desktop”

Object-Based Languages

Language object-based if supports objects as feature

Object-based languages:

Ada, Modula, Clu, C++, Java, Scala, C#, Smalltalk, and Python 3

Not inherently object-based:

Pascal (without module extensions), Algol, Fortran, C

Model: Classes (1)

Class – template or factory for creating objects

• describes collection of related objects – instances

• common operations and possible states for instances

• closely related to concept of type

Model: Classes (2)

Class description includes definitions of:

• operations

• set of possible states

Classroom Sim Classes

• Class StudentDesk

  • appropriate data attributes and operations

  • template for instances

• Other classes

Class-based Languages

Class-based language

• object-based language with “class” as feature

• every object has class

Class-based languages:

Clu, C++, Java, Scala, C#, Smalltalk, Ruby, Ada 95:

Not class-based:

Ada 83, Modula, Lua, core JavaScript,

Classes and Types

• Some statically typed languages treat classes as (nominal) types: Java, Scala, C++, C#

• Some dynamically types languages treat types as what operations supported (duck typing): Smalltalk, Ruby

• Python 3? both nominal typing and duck typing

• Types chapter has more detail

Model: Inheritance (1)

Goal: Encourage reuse of design and code

Class C inherits from class P if C’s objects form a subset of P’s objects.

• Class C supports all class P operations/state

• Class C may support more operations/state

• Class C subclass, child, derived class

• Class P superclass, parent, base class

• Class P generalization of class C; C specialization of P

Classroom Sim Inheritance (1)

    // Java or Scala
    class Furniture  // extends cosmic root class for references
    {   ...   }      // (java.lang.Object, scala.AnyRef)

    class Desk extends Furniture
    {   ...   }  // implement "move" op

    class StudentDesk extends Desk
    {   ...   }  // share "move" from Desk

    class ComputerDesk extends Desk
    {   ...   } // share "move" from Desk

Classroom Sim Inheritance (2)

Figure 3-1: Classroom Simulation Inheritance Hierarchy

Classroom Sim Inheritance (3)

Figure 3-2: Classroom Simulation with Multiple Inheritance

Classroom Sim Inheritance (4)

Problems with multiple inheritance in StudentDesk

• What if “move” in both Desk and Chair

• What if data fields in Furniture – 1 copy or 2?

Solutions complex:

• Java and Scala – do not allow – use interfaces

• Other languages allow (including C++ and Python 3) but have complex rules

Classroom Sim Inheritance (5)

    // Java
    interface Location3D
    {   ...   } // variable defs, method signatures

    interface HumanHolder
    {   ...   }

    interface Seat extends Location3D, HumanHolder
    {   ...   } // default method defs in newer Java

    interface BookHolder
    {   ...   }

    interface BookBasket extends Location3D, BookHolder
    {   ...   }

    class StudentDesk extends Desk implements Seat, BookBasket
    {   ...   }

Classroom Sim Inheritance (6)

Figure 3-3: Classroom Simulation with Interfaces

Model: Subtype Polymorphism (1)

Polymorphism – “many forms”:

hide different implementations behind common interface

Subtype polymorphism:

associate operation call with appropriate implementation in class hierarchy

also called polymorphism by inheritance, inclusion polymorphism, subtyping

Usually at runtime – dynamic binding to nearest implementation

Model: Subtype Polymorphism (2)

    class Desk extends Furniture
    {   ...
        public void move(...)
        ...
    }

    class StudentDesk extends Desk
    {   ...
        // no move(...) operation here
        ...
    }

    class ComputerDesk extends Desk
    {   ...
        // no move(...) operation here
        ...
    }

Model: Subtype Polymorphism (3)

    class Desk extends Furniture 
    {   ...
        public void move(...)
        ...
    }

    class StudentDesk extends Desk
    {   ...
        // no move(...) operation here
        ...
    }

    class ComputerDesk extends Desk
    {   ... // what about cords?
        public void move(...)
        ...
    }

Python 3 Class (1)

    class CountingOO:                  # (1)

        def __init__(self,c,m):        # (2,3)
            self.count = c             # (4)
            self.maxc  = m

        def has_more(self,c,m):        # (5)
            return c <= m 

        def adv(self):                 # (6)
            self.count = self.count + 1
    
        def counter(self):             # (7)
            while self.has_more(self.count,self.maxc):
                print(f'{self.count}') # (8)
                self.adv()

Python 3 Class (2)

    class CountingOO:                  # (1)

        def __init__(self,c,m):        # (2,3)
            self.count = c             # (4)
            self.maxc  = m

1. Inherits from cosmic root class object by default – otherwise parents in parenthesis after class name

2. Uses initializer __init__, not really constructor

3. Lists receiver object as first parameter – convention self

4. Has instance variables count, maxc; access with self.

Python 3 Class (3)

        def has_more(self,c,m):        # (5)
            return c <= m 

        def adv(self):                 # (6)
            self.count = self.count + 1
    
        def counter(self):             # (7)
            while self.has_more(self.count,self.maxc):
                print(f'{self.count}') # (8)
                self.adv()

5. Boolean function has_more() – hook method in Template Method design pattern

6. Procedure adv() sets self.count – hook method

Python 3 Class (4)

        def counter(self):             # (7)
            while self.has_more(self.count,self.maxc):
                print(f'{self.count}') # (8)
                self.adv()

7. Procedure counter() – primary public interface to object – template method in Template Method design pattern

8. Expression f'{self.count}' – Python 3.7 interpolated string

Python 3 Class (5)

    class CountingOO:                  # (1)

        def __init__(self,c,m):        # (2,3)
            self.count = c             # (4)
            self.maxc  = m

        def has_more(self,c,m):        # (5)
            return c <= m 

        def adv(self):                 # (6)
            self.count = self.count + 1
    
        def counter(self):             # (7)
            while self.has_more(self.count,self.maxc):
                print(f'{self.count}') # (8)
                self.adv()

Using Python 3 Class

    ctr = CountingOO(0,10)
    ctr.counter()

Gives output

    0
    1
    2
    ...
    10

Extending Python 3 Class (1)

    class Times2(CountingOO):    # inherits from CountingOO

        def has_more(self,c,m):  # overrides
            return c != 0 and abs(c) <= abs(m)

        def adv(self):           # overrides
            self.count = self.count * 2

Extending Python 3 Class (2)

    ctr2 = Times2(-1,10)
    ctr2.counter()

Gives output

    -1
    -2
    -4
    -8

Prototype-based Paradigm

Prototype-based language

• Has objects but no classes

• Copies (clones) existing object (prototype) to create new

Prototype:

• Object with collection of slots

• Each slot data attribute or operation.

Classes vs. Prototypes

• Class-based language have crisp boundaries – in class or not

• Prototype-based language have fuzzy boundaries when needed

• Categories blend into each other

• Cloning approach more flexible

Student Who Plays Chess (1)

Using classes

• Have Student class

• Want some Student to be chess player

• Create new class ChessPlayingStudent

  But inherit from Student or ChessPlayer?

  Or compose Student and ChessPlayer (have as fields of ChessPlayingStudent)

Student Who Plays Chess (2)

Using prototype objects

• Have Student object

• Want some Student to be chess player

• Clone existing object, dynamically add new data attributes and operations – get ChessPlayingStudent object

Evolution of Tony

• Consider Student named Tony

• Graduates with BSCS becomes Alumnus

• But hires in IT, also becomes Staff

• Joins graduate CS program, so still Student

• Teaches course, also becomes Faculty

• Learns to play chess

Can evolve object Tony object using prototype cloning

Delegation

May delegate to prototype object rather then copy it completely

• Put new or changed attributes in new object slots

• Delegate unchanged data attribues and operations to prototype

Prototypes Into Classes

Prototype and delegation

• more basic than classes and inheritance

• can simulate classes and inheritance

Prototype-Based Languages

• Self

• NewtonScript, JavaScript, Lua, Io

• Python (with package prototype.py)

Lua Design Principles

Dynamically typed, multiparadigm language with design principles:

• portability
• embeddability
• efficiency
• simplicity

Lua Implementation Requirements

• Can only use standard C language and library

• Must be efficient in use of memory and processor time

• Must support interoperability with C programs in both directions

Why?

• C ubiquitous across hardware/OS platforms

• Gives portability, embeddability, efficiency

Lua Usage

• Extending apps with user-written Lua scripts

• In game scripting, Wikipedia templates, Pandoc filters, etc.

Separate Mechanism from Policy

• Provide meta-mechanisms for implementing features instead of lots of features

• Makes Lua simple but powerful

Examples:

• No classes or inheritance, but mechanism to build them

• Table only data structure, but flexible and efficient

• Metatables and metamethods

• First class functions (closures)

Tables Are Objects

• State (i.e. values associated with keys)

• Operations by associating function closures with keys

  Closure is function def plus free variable values

• Identity independent of state

• Lifecycle independent of the code that created it

• No direct encapsulation, rely on convention and advanced techniques

Method Call

Reference to function closure in table obj (expression) at key method

    obj.method(obj, other_arguments)

Represented by syntactic sugar

    obj:method(other_arguments) 

Receiver obj called parameter self in body

Delegation

Metatable mechanism with metamethod __index:

• delegates missing key access to different table

• enables prototype-based paradigm

Lua Prototype Module (1)

    -- File CountingPB.lua 
    local CountingPB = {count = 1, maxc = 0} -- (1)
    
    function CountingPB:new(mixin)           -- (2)
       mixin = mixin or {}                   -- (5)
       local obj = { __index = self }        -- (4)
       for k, v in pairs(mixin) do           -- (5)
          if k ~= "__index" then
             obj[k] = v
          end
       end
       return setmetatable(obj,obj)          -- (6,7)
    end

    function CountingPB:has_more(c,m)        -- (2)
       return c <= m
    end
    
    function CountingPB:adv()                -- (2)
       self.count = self.count + 1
    end
    
    function CountingPB:counter()            -- (2)
       while self:has_more(self.count,self.maxc) do
          print(self.count)
          self:adv()
        end
    end
    
    return CountingPB                        -- (3)

Lua Prototype Module (2)

    -- File CountingPB.lua 
    local CountingPB = {count = 1, maxc = 0} -- (1)
       ...
    function CountingPB:new(mixin)           -- (2)
       ...
    function CountingPB:has_more(c,m)        -- (2) 
       ...
    function CountingPB:adv()                -- (2)
       ...
    function CountingPB:counter()            -- (2)
       ...    
    return CountingPB                        -- (3)

1. Create module CountingPB as table with default values (prototype obj)

2. Define methods new(), has_more, etc

3. Return CountingPB from require call

Lua Prototype Module (3)

    function CountingPB:new(mixin)           -- (2)
       mixin = mixin or {}                   -- (5)
       local obj = { __index = self }        -- (4)
       for k, v in pairs(mixin) do           -- (5)
          if k ~= "__index" then
             obj[k] = v
          end
       end
       return setmetatable(obj,obj)          -- (6,7)
    end

Method new (arbitrary name) constructs clones

4. Creates as table with __index set to receiver

5. Copies mixin entries into clone, adds or changes

6. Sets clone’s metatable to clone itself

7. Returns clone object – module convention

Lua Prototype Module (4)

On access to undefined key

• If no metatable set, return nil

• If metatable set, check __index entry

  • if table, then delegate access to that
  • if function closure, then call it
  • else return nil

Prototype delegation functionality

Lua Prototype Module (5)

    -- File CountingPB.lua 
    local CountingPB = {count = 1, maxc = 0} -- (1)
    
    function CountingPB:new(mixin)           -- (2)
       mixin = mixin or {}                   -- (5)
       local obj = { __index = self }        -- (4)
       for k, v in pairs(mixin) do           -- (5)
          if k ~= "__index" then
             obj[k] = v
          end
       end
       return setmetatable(obj,obj)          -- (6,7)
    end

    function CountingPB:has_more(c,m)        -- (2)
       return c <= m
    end

    function CountingPB:adv()                -- (2)
       self.count = self.count + 1
    end

    function CountingPB:counter()            -- (2)
       while self:has_more(self.count,self.maxc) do
          print(self.count)
          self:adv()
        end
    end
    
    return CountingPB                        -- (3)

Using Lua Prototype Module (1)

Load CountngPB.lua module, create object, call counter

    local CountingPB = require "CountingPB"

    x = CountingPB:new({count = 0, maxc = 10})
    x:counter()

Gives output

    0
    1
    2
    ...
    10

Using Lua Prototype Module (2)

Create clone of object x

    y = x:new({count = 10, maxc = 15})
    y:counter()

Give output

    10
    11
    12
    13
    14
    15

Using Lua Prototype Module (3)

    z = y:new( { maxc = 400,
                 has_more = function (self,c,m)
                    return c ~= 0 and math.abs(c) <= math.abs(m)
                 end,
                 adv = function(self)
                    self.count = self.count * 2
                 end,
                 bye = function(self) print(self.msg) end 
                 msg = "Good-Bye!" } )
    z:counter()
    z:bye()

Gives output

    16
    32
    64
    128
    256
    Good-Bye!

OO versus PB

• OO paradigm usually enforces particular policy for object creation and inheritance

  Programming outside policy difficult

• PB paradigm provides low-level mechanisms – allows programmers to define own policies flexibility in code

  Can give incompatible structures from one program to another

Key Ideas

• Defined object characteristics

  Essential: state, operations, identity
  Important: encapsulation; independent lifecycle
  

• Defined general object model for OO paradigm – objects, classes, inheritance, subtype polymorphism

• Examined Python 3 OO class example

• Defined object-based, class-based, and prototype-based paradigms

Source Code

• Python 3 example: CountingOO.py

  Similar OO Scala example: CountingOO.scala

• Lua example: CountingPB.lua

  Example use CountingPB_Test.lua