Pure Fabrication and

“Gang of Four”
Design Patterns
Larman, chapters 25 and 26
CSE432
Object-Oriented Software Engineering
Glenn D. Blank, Lehigh University
Pure Fabrication
Another GRASP design pattern

Problem: you must assign a responsibility to a class,

but assigning it to a class that already represents
a problem domain entity would ruin its low
coupling and/or high cohesion.
Solution: Assign a highly cohesive set of
responsibilities to “made up,” fabricated class—it does
not need to represent a problem domain concept—in
order to support high cohesion, low coupling & reuse.
Pure Fabrication: Example
 For NextGen POS, there is a requirement to save the Sale
instances in a database.
 What does Information Expert tell us to do?
 Assign this responsibility to the Sale class,
since Sale has the data that needs to be saved.
 But the above solution leads to:
 Low Cohesion: database tasks not related to Sale High Coupling:
Should Sale interface with database?
 Low Reusability: General task of saving to a database is one that
other classes may share.
Instead, fabricate a new class,
PersistentStorage
Invent a new class that is solely responsible
for saving objects.
Pure Fabrication Benefits
 High Cohesion: fabricated class focuses on a
very specific responsibility
 Reuse: fine-grained pure fabrication classes
with specific responsibilities are relatively
easy to understand and reuse in other
applications
 Pure fabrication principle leads to many other
reusable design patterns, including most of
the Gang of Four patterns
1995
The “gang of four” (GoF)
Design Patterns book catalogs 23 different patterns
 Solutions to different classes of problems, in C++ & Smalltalk

 Problems and solutions are broadly applicable, used by many

people over many years
 Patterns suggest opportunities for reuse in analysis,
design and programming
 GOF presents each pattern in a structured format
 What do you think of this format? Pros and cons?
Elements of Design Patterns
 Design patterns have 4 essential elements:
 Pattern name: increases vocabulary of designers
 Problem: intent, context, when to apply
 Solution: UML-like structure, abstract code
 Consequences: results and tradeoffs
Design Patterns are NOT
 Data structures that can be encoded in classes and
reused as is (i.e., linked lists, hash tables)
 Complex domain-specific designs
(for an entire application or subsystem)
 If they are not familiar data structures or complex domain-
specific subsystems, what are they?

 They are:
 “Descriptions of communicating objects and classes
that are customized to solve a general design problem in a
particular context.”
Three Types of GoF Patterns
 Creational patterns:
 Deal with initializing and configuring objects
 Structural patterns:
 Composition of classes or objects
 Decouple interface and implementation of classes
 Behavioral patterns:
 Deal with dynamic interactions among societies of objects
 How they distribute responsibility
Structural patterns
 Assemble objects to realize new functionality
 Exploit flexibility of object composition at run-time
 Not possible with static class composition
 Example: Proxy
 Proxy acts as convenient surrogate or placeholder
for another object.
 Examples?
 Remote Proxy: local representative for object in a
different address space
 Virtual Proxy: represent large object that should be
loaded on demand
 Protected Proxy: protect access to the original object
Structural Patterns
 Adapter:
 Converts interface of a class into one that clients expect
 Bridge:
 Links abstraction with many possible implementations
 Composite:
 Represents part-whole hierarchies as tree structures
 Decorator:
 Attach additional responsibilities to object dynamically
 Facade:
 Simplifies the interface for a subsystem
 Flyweight:
 Shares many fine-grained objects efficiently
 Proxy:
 Provides a surrogate or placeholder for another object to control
access to it
Adapter pattern
Problem: How to resolve incompatible interfaces or
provide a stable interface to similar components
with different interfaces?
Solution: Convert original interface component into
another one through an intermediate adapter.

Use interfaces and polymorphism to add indirection

to varying APIs
POS example: Instantiate
adapters for external services
«interface» Adapters use interfaces and
ITaxCalculatorAdapter polymorphism to add a level of
indirection to varying APIs in other
getTaxes( Sale ) : List of TaxLineItems components.

TaxMasterAdapter GoodAsGoldTaxPro
Adapter

getTaxes( Sale ) : List of TaxLineItems

getTaxes( Sale ) : List of TaxLineItems

«interface» «interface»
IAccountingAdapter ICreditAuthorizationService
Adapter
postReceivable( CreditPayment )
postSale( Sale ) requestApproval(CreditPayment,TerminalID, MerchantID)
... ...

«interface»
IInventoryAdapter
SAPAccountingAdapter GreatNorthernAccountingAdapter
...

postReceivable( CreditPayment ) postReceivable( CreditPayment )

postSale( Sale ) postSale( Sale )
... ...
Fig. 26.1
Using an Adapter: adapt postSale
request to SOAP XML interface

:Register : SAPAccountingAdapter

makePayment
...
SOAP over
HTTP

postSale( sale )
xxx «actor»
: SAPSystem

the Adapter adapts to

interfaces in other components

Fig. 26.2
Benefits of Adapter pattern
 Reduces coupling to implementation specific details
 Polymorphism and Indirection reveals essential
behavior provided
 Including name of design pattern in new class (e.g.,
TaxMasterAdapter) in class diagrams and code
communicates to other developers in terms of
known design patterns
Creational Patterns
 Singleton: Guarantee access to a singular (sole) instance
 Simple Factory: Create specialized, complex objects
 Abstract Factory: Create a family of specialized factories
 Factory Method: Define an interface for creating an
object, but let subclasses decide which class to instantiate
 Builder: Construct a complex object step by step
 Prototype: Clone new instances from a prototype
 Lazy initialization: Delay costly creation until it is needed
Singleton pattern (creational)
 A class with just instance and provide a global point of access
 Global Variables can be dangerous!
(side effects, break information hiding)

class Singleton
{ public:
static Singleton* getInstance();
protected: //Why are the following protected?
Singleton();
Singleton(const Singleton&);
Singleton& operator= (const Singleton&);
private: static Singleton* instance;
};
Singleton *p2 = p1->getInstance();
Simple Factory pattern
 Context/Problem
 Who should be responsible for creating objects
when there are special considerations, such as
complex logic,a desire to separate the creation
responsibilities for better cohesion, and so forth
 Solution
 Create a Pure Fabrication to handle the creation
Factory can create different
objects from a file
ServicesFactory note that the factory methods
return objects typed to an
accountingAdapter : IAccountingAdapter interface rather than a class, so
inventoryAdapter : IInventoryAdapter that the factory can return any
taxCalculatorAdapter : ITaxCalculatorAdapter implementation of the interface

getAccountingAdapter() : IAccountingAdapter
getInventoryAdapter() : IInventoryAdapter
getTaxCalculatorAdapter() : ITaxCalculatorAdapter
...

if ( taxCalculatorAdapter == null )
{
// a reflective or data-driven approach to finding the right class: read it from an
// external property

String className = System.getProperty( "taxcalculator.class.name" );

taxCalculatorAdapter = (ITaxCalculatorAdapter) Class.forName( className ).newInstance();

}
return taxCalculatorAdapter;

Figure 26.5
Advantages of Factory Objects?
 Separates responsibility of complex creation into
cohesive helper classes
 Hides complex creation logic, such as
initialization from a file
 Handles memory management strategies,
such or recycling or caching
Use Singleton to create a Factory
UML notation: this '1' can optionally be used to
indicate that only one instance will be created (a
singleton)

1
ServicesFactory

UML notation: in a instance : ServicesFactory singleton static

class box, an attribute
underlined attribute or accountingAdapter : IAccountingAdapter
method indicates a inventoryAdapter : IInventoryAdapter
static (class level) taxCalculatorAdapter : ITaxCalculatorAdapter
member, rather than singleton
an instance member getInstance() : ServicesFactory static
method
getAccountingAdapter() : IAccountingAdapter
getInventoryAdapter() : IInventoryAdapter
getTaxCalculatorAdapter() : ITaxCalculatorAdapter
...

// static method
public static synchronized ServicesFactory getInstance()
{
if ( instance == null )
instance = new ServicesFactory()
return instance
} Figure 26.6
Adapter, Factory and Singleton
working together
1
:Store
:ServicesFactory

create
create :Register

accountingAdapter =
getAccountingAdapter
create : SAPAccounting
Adapter

accountingAdapter:
:Register
SAPAccountingAdapter

makePayment
create(cashTendered) : Payment SOAP over
HTTP

postSale( sale )
xxx «actor»
: SAPSystem

Figure 26.8
Behavioral Patterns
 Chain of Responsibility:
 Request delegated to the responsible service provider
 Command:
 Request or Action is first-class object, hence storable
 Iterator:
 Aggregate and access elements sequentially
 Interpreter:
 Language interpreter for a small grammar
 Mediator:
 Coordinates interactions between its associates
 Memento:
 Snapshot captures and restores object states privately

Which ones do you think you have seen somewhere?

Behavioral Patterns (cont.)
 Observer:
 Observers update automatically when observed
object changes
 State:
 Object whose behavior depends on its state
 Strategy:
 Abstraction for selecting one of many algorithms
 Template Method:
 Algorithm with some steps supplied by derived class
 Visitor:
 Operations applied to elements of a heterogeneous
object structure
Strategy design pattern
Problem: How to design a family of algorithms or policies
that are essentially the same but vary in details?
Solution: "Define a family of algorithms, encapsulate each
one, and make them interchangeable." [Gamma, p315]
 Use abstraction and polymorphism to show high level
algorithm and hide varying implementation details
 Multiple SalePricingStrategy classes
with polymorphic getTotal method
«interface»
ISalePricingStrategy

getTotal( Sale ) : Money

PercentDiscount AbsoluteDiscount ???

PricingStrategy OverThreshold PricingStrategy
PricingStrategy
percentage : float ...
discount : Money
getTotal( s:Sale ) : threshold : Money ...
Money
getTotal( s:Sale ) :
Money

{ {
return s.getPreDiscountTotal() * percentage pdt := s.getPreDiscountTotal()
} if ( pdt < threshold )
return pdt
else
return pdt - discount Figure 26.9
}
Observer pattern
 Intent:
 Define a one-to-many dependency between objects
so that when one object changes state, all its dependents
are notified and updated automatically
 Used in Model-View-Controller framework
 Model is problem domain
 View is windowing system
 Controller is mouse/keyboard control
 How can Observer pattern be used in other
applications?
 JDK’s Abstract Window Toolkit (listeners)
 Java’s Thread monitors, notify(), etc.
Structure of Observer Pattern
Subject Observer

for all observers obs +notify() * +update()

{
obs->update()
+attach(in Observer) 1

} +detach(in Observer)

ConcreteObserver
ConcreteSubject
-subjectSate * +update()
+getState() 1

return subjectState observerState = subject->getState()

Patterns in software libraries
 AWT and Swing use Observer pattern
 Iterator pattern in C++ template library & JDK
 Façade pattern used in many student-
oriented libraries to simplify more
complicated libraries!
 Bridge and other patterns recurs in
middleware for distributed computing
frameworks
 …
Command pattern
 Synopsis or Intent: Encapsulate a request as an object,
thereby letting you parameterize clients with different requests,
queue or log requests, and support undoable operations
 Context: You want to model the time evolution of a program:
 What needs to be done, e.g. queued requests, alarms, conditions
for action
 What is being done, e.g. which parts of a composite or distributed
action have been completed
 What has been done, e.g. a log of undoable operations
 What are some applications that need to support undo?
 Editor, calculator, database with transactions
 Perform an execute at one time, undo at a different time
 Solution: represent units of work as Command objects
 Interface of a Command object can be a simple execute() method
 Extra methods can support undo and redo
 Commands can be persistent and globally accessible, just like
normal objects
 Command pattern, continued
 Structure:

Participants (the classes and/or objects participating in this pattern):

Command  (Command) declares an interface for executing an operation
ConcreteCommand  defines a binding between a Receiver object and an action
implements Execute by invoking the corresponding operation(s) on Receiver
Invoker asks the command to carry out the request
Receiver knows how to perform operations associated with carrying out the request
Client creates a ConcreteCommand object and sets its receiver
Command pattern, continued
 Consequences:
 You can undo/redo any Command
 Each Command stores what it needs to restore state
 You can store Commands in a stack or queue
 Command processor pattern maintains a history
 It is easy to add new Commands, because you do
not have to change existing classes
 Command is an abstract class, from which you derive
new classes
 execute(), undo() and redo() are polymorphic functions
 More software patterns
 Language idioms (low level, C++): Jim Coplein, Scott Meyers
 I.e., when should you define a virtual destructor?
 Architectural (systems design): layers, reflection, broker
 Reflection makes classes self-aware, their structure and
behavior accessible for adaptation and change:
Meta-level provides self-representation, base level
defines the application logic
 Java Enterprise Design Patterns (distributed transactions
and databases)
 E.g., ACID Transaction: Atomicity (restoring an object after a
failed transaction), Consistency, Isolation, and Durability
 Analysis patterns (recurring & reusable analysis models, from
various domains, i.e., accounting, financial trading, health)
 Process patterns (software process & organization)
Benefits of Design Patterns
 Design patterns enable large-scale reuse of
software architectures and also help document
systems
 Patterns explicitly capture expert knowledge and
design tradeoffs and make it more widely available
 Patterns help improve developer communication
 Pattern names form a common vocabulary
Web Resources
 http://home.earthlink.net/~huston2/dp/
 http://www.dofactory.com/
 http://hillside.net/patterns/