Refactoring AOMs For AgilePT2010

Agile Refactoring for Making
Systems More Adaptable
Joseph W. Yoder
The Refactory, Inc.
June 25, 2010
joe@refactory.com
http://www.refactory.com

The Refactory Principals
John Brant Don Roberts
Brian Foote Joe Yoder
Ralph Johnson

Refactory Affiliates
Joseph Bergin
Fred Grossman
Bill Opdyke
Slide - 2
Agile Refactoring for Making Systems More Adaptable Copyright 2010, Joseph W. Yoder & The Refactory, Inc.

Agile to the Rescue???
Individuals and interactions over processes and tools
Working software over comprehensive documentation
Customer collaboration over contract negotiation
Responding to change over following a plan

That is, while there is value in the items on
the right, we value the items on the left more.

…From the Agile Manifesto
Slide - 3

Agile Principles
Scrum, TDD, Refactoring, Regular Feedback,
Testing, More Eyes, …
Good People! Face-To-Face conversation.
Continuous attention to technical excellence!
Motivated individuals with the environment
and support they need. Retrospectives!
Allow Requirements to Change! Encourage
Software Evolution as needed!
Slide - 4

Agile Encourages Changes
Very Little Upfront Design!
Piecemeal Growth! Small Iterations!
Late changes to the requirements
of the system!
Continuously Evolving the Architecture!

Adapting to Changes requires the code to
change and Refactoring supports changes
to the code in a “safe” way.
Slide - 5

Agile & Refactoring
A key part of Agile is to allow things to
change and adapt as dictated by
business needs!
To support these changes, Refactoring is
encouraged by most Agile practitioners.
Agile has helped Refactoring be accepted
into the mainstream development
process (even sometimes encouraged).
Slide - 6

Ground Breaking Work
Before there was Agile…

Started at the University Of Illinois

Bill Opdyke & Don Roberts PHD Thesis
with Professor Ralph Johnson

Refactoring Browser by John Brant and
Don Roberts (first commercial tool)
Slide - 7

Refactoring Definition
Refactoring is the process of changing a
software system in such a way that it
does not alter the external behavior of
the code, yet improves its internal
structure…Martin Fowler, Refactoring
Improving the Design of Existing Code;
Addison Wesley , 1999

Slide - 8

What is Refactoring
Refactoring is a controlled technique for improving the
design of an existing code base

Small behavior-preserving transformations,
• Each of might be "too small to be worth doing“
• Cumulative effect of transformations is significant
• Small steps help reduce the risk of introducing errors
• Avoid having the system broken while you are
carrying out the restructuring
• Gradually refactor a system over a period of time

Slide - 9

A Simple Refactoring
Create Empty Class
Object

Object

NewAbstract

Concrete1 Concrete2

Concrete1 Concrete2

Slide - 10

A Complex Refactoring

Array
MatrixRep

Matrix
rep

Matrix

SparseRep ArrayRep IdentityRep

Adapted from Don Roberts, The Refactory, Inc. Slide - 11

Refactoring Code Smells
A code smell is a hint that something has gone wrong
somewhere in your code. Use the smell to track
down the problem. KentBeck (with inspiration from
the nose of Massimo Arnoldi) seems to have coined
the phrase in the "OnceAndOnlyOnce" page, where
he also said that code "wants to be simple". Bad
Smells in Code was an essay by KentBeck and
MartinFowler, published as Chapter 3 of Refactoring
Improving The Design Of ExistingCode.
----Wards Wiki

Slide - 12

Code Smells Review
Duplicate Code, Long Method, Large Class,
Long Parameter List, Divergent Change,
Shotgun Surgery, Feature Envy, Data Clumps
Primitive Obsession, Switch Statements
Parallel Inheritance, Lazy Class
Speculative Generality, Temporary Field
Message Chains, Middle Man
Inappropriate Intimacy, Incomplete Lib Classes
Data Class, Refused Bequest, Comments
Slide - 13

Comments
We are not against comments but…

If you see large methods that have
places where the code is commented,
use Extract Method to pull that out to a
comment

Slide - 14

Comments Example
(Extract Method)
void printOwing (double amount) {
printBanner();
//print details
System.out.println (name: “ + _name);
System.out.println (amount: “ + amount);
…}

void printOwing (double amount) {
printBanner();
printDetails();
…}
void printDetails (double amount) {
System.out.println (name: “ + _name);
System.out.println (amount: “ + amount);} Slide - 15

Extract Method Mechanics
Create a new method and name it after the intention of
the code to extract.
Copy the extracted code from the source to the new
method.
Scan the extracted code for references to any variables
that are local to the source method.
See whether any temps are used only within extracted
code.
Pass into target method as parameters local-scope
variables that are read from the extracted code.
Replace the extracted code in the source method.

Slide - 16

Remove Method with
Method Object
Class Order …
double price() {
double primaryBasePrice;
double secondaryBasePrice;
double tertiaryBasePrice;
// long computation;
…}

Order PriceCalculator
price() primaryBasePrice
1 secondaryBasePrice
tertiaryBasePrice

return new PriceCalculator(this).compute() compute()

Slide - 17

Remove Method to
Method Object Mechanics
Create a new class named after method.
Give the new class a final field for the object
that hosted the original method.
Give the new class a constructor for the original
object and each parameter.
Give the new class a compute method.
Copy the body of original method to compute.
Compile.
Replace the old method with the one that
creates the new object and calls compute.
Slide - 18

Agile & Testing
Many Agile practices highly encourage
Testing as one of their core practices.
Processes like XP support developers
writing many unit tests (XUnit ).
Test Driven Development (TDD) is
becoming highly accepted in Agile.
Testing was a key principle of Refactoring
before there was Agile!
Slide - 19

Prerequisites of Refactoring
Since you are changing the code base,
it is IMPORTANT to validate with tests.

There are also a time to refactor and a
time to wait.

Need both Unit Testing and Integrated
Tests for making sure nothing breaks

Slide - 20

Deciding Whether A
Refactoring is Safe
"A refactoring shouldn't break a program.”
 What does this mean?
A safe refactoring is behavior preserving.
It is important not to violate:
 naming/ scoping rules.
 type rules.
"The program should perform the same
after a refactoring as before.”
Satisfying timing constraints.
Slide - 21

Using Standard Tools
To be safe, must have tests
Should pass the tests before
and after refactoring
 Commercial Testing Tools
 Kent Beck’s Testing Framework
(SUnit, JUnit, Nunit, …)
Take small steps, testing between each
Java and C# tools are getting better

Slide - 22

Catalogue of Refactorings
Composing Method
Moving Features
Organize Data
Simplifying Conditionals
Simpler Method Calls
Generalization

From Fowlers Book
Slide - 23

Refactoring and
Design Patterns (1)

What varies Design Pattern
Algorithms Strategy, Visitor
Actions Command
Implementations Bridge
Response to change Observer

Slide - 24

Refactoring and
Design Patterns (2)
What varies Design Pattern
Interactions between objects Mediator
Factory Method, Abstract
Object being created Factory, Prototype
Structure being created Builder
Traversal Algorithm Iterator
Object interfaces Adapter
Object behavior Decorator, State
Slide - 25

Refactoring Summary
We have seen the mechanics for
some of the Refactorings.

When you find smelly code, you often
apply Refactorings to clean your code.

Refactorings do often apply
Design Patterns.
Slide - 26

Building a System to Change
Sometimes it pays off to build a system that
can respond quickly to change!

Slide - 27

Allowing Systems to Adapt
Requirements change within
application’s domain.
Domain Experts know their domain best.
Business Rules are changing rapidly.
Applications have to quickly adapt to
new business requirements.
Agile Development Welcomes Changes to the
Requirements, even late in the development!
One way to Adapt is to just Refactor to new
business requirements…though sometimes…
Slide - 28

Forces – Shearing Layers

Who (Business Person, Analyst, Developer)
What (Business Rule, Persistence Layer,…)
When (How often, How fast)

There is a different rate of change
on the system.
Foote & Yoder - Ball of Mud PLoPD4
Slide - 29

Newborn Screening
Refactoring Example

Doctor, HealthProf.
Hospital, Lab

Blood Specimen
Mother, Infant

Slide - 30

Medical Observation –
Basic OO Design Model
Person * Observation

Measurement Trait
Quantity
traitValue
unit convertTo:
amount
expressOnUnit:
expressOnUnits:

PhysicalMeasure Blood
EyeColor HairColor Gender …

Height Weight FeedingType Hearing Vision
…

What happens when a new observation is required?
Slide - 31

Observation Design
(1st Design)
Person Observation
ObservationType
1 * -recordedDate : Date * 1
+address() -observedDate : Date -phenomenon : Symbol
+phone() -duration : TimePeriod
+name() +phenomenon()

Quantity
Measurement Trait
1 1
-unit : Unit
-amount : Number
+expressOnUnit(aUnit : Unit) +observationValue() +observationValue()
+expressOnUnits(unitCollection : Collection)

Slide - 32

Observation Design
Example

Height: 3 feet

Eyes Color: Blue
Name: John Smith
Mother: Sue Smith
Father:
Address:
Phone:

Slide - 33

Observation Design
(instance diagram)
aPerson
name <John Smith>
obsCollection

anObservationType
aMeasurement
#height
type
value

aQuantity
value <3>
unit <ft>

aTrait
type anObservationType
value <blue> #eyeColor

Slide - 34

Composing Observations
Observations can be more complex
Cholesterol
 Components: HDL, LDL
Blood Pressure
 Components: Systolic, Diastolic
Vision
 Components: Left Eye, Right Eye
Slide - 35

Composite Observation Design
(1st Refactoring)
Person

+address()
+phone()
+name()

Observation
ObservationType
-recordedDate : Date
-phenomenon : Symbol -observedDate : Date 1..n
-duration : TimePeriod
+phenomenon()

Quantity
Measurement Trait CompositeObservation
-unit : Unit
-amount : Number -observations : Collection
+expressOnUnit(aUnit : Unit) +observationValue() +observationValue()
+expressOnUnits(unitCollection : Collection)

Composite Pattern (GOF)
Make sure all tests still pass! Slide - 36

Observation Design
Example

Height: 3 feet

Eyes Color: Blue
Name: John Smith
Mother: Sue Smith Blood Pressure:
Father: Systolic: 120 mmHg
Address: Diastolic: 80 mmHg
Phone:

Slide - 37

Composite Observation Design
(instance diagram)
anObservType
<#BloodPressure>
aCompObs

aMeasurement anObserType
<120 mmHg> <#SYSTOLIC>

anotherMeasurement anObser-Type
<80 mmHg> <#DIASTOLIC>

Slide - 38

Composite and Primitive
Observation Design
What we know about John?
aPerson
name <John Smith>
obsCollection

EyesColor: Blue Height: 3 feet BloodPressure: 120/80

Systolic: 120 mmHg

Diastolic: 80 mmHg

Slide - 39

Validating Observations

Each Observation has its own set
of legal values.
 Baby’s Weight: [0 30] pounds
 HepatitisB: {positive, negative}
 Left/Right Vision: {normal, abnormal}
GUI can enforce legal values
 but we want business rules in domain
objects

Slide - 40

Validating Observations Design
(2nd Refactoring)
ObservationType
Validator

-phenomenon : Symbol
-validator : Validator

DiscreteValidator NullValidator RangedValidator

-descriptorSet : Collection -intervalSet : Collection
-validUnit : Unit

Slide - 41

Overall Observation Design
Validator

DiscreteValidator RangedValidator NullValidator
Party descriptorSet intervalSet
validUnit

Observation
ObservationType recordedDate
type comments

Measurement Trait CompositeObservation
value value values
value value
value: value:
convertTo:

Quantity
First make sure original test cases pass and
then add new test cases for the validators!
Is everything an Observation?
How does the model specify the structure of the Composite?
What is the relationship between Trait and DiscreteValidator?
Slide - 42

Observation Design
Validator
validatorName
ObservationType isValid: obsValue
phenomenonType
isValid: obsValue
descriptorSet intervalSet
validUnit
PrimitiveObservation CompositeObservation
Type Type Party

Observation
recordedDate
comments
isValid
Quantity
unit
CompositeObservation Primitive Observation quantity
observationValue convertTo:

DiscreteValues

Extend the Design by adding Composite to Type
-- Refactor the Metadata
Slide - 43

Observation Design
Validator
validatorName
ObservationType isValid: obsValue
phenomenonType
isValid: obsValue
validUnit
Type Type Party

Observation
recordedDate
comments
isValid
Quantity
unit

Operational level DiscreteValues

Slide - 44

Observation Design
Validator
validatorName
ObservationType Knowledge level isValid: obsValue
phenomenonType
isValid: obsValue
validUnit
Type Type Party

Observation
recordedDate
comments
isValid
Quantity
unit

DiscreteValues

Slide - 45

Observation Design
(instance diagram)

anObserv-Type
<#COMP-GAL>
aCompObs

aMeasurement anObser-Type aRangedValidator
<aQuantity> <#GAL>

anotherMeasurement anObser-Type anotherRangedValidator
<anotherQuantity> <#UDT>

Slide - 46

Observation Design
(instance diagram)
anInfant
name aDiscreteValidator
obsCollection
anObservationType aRangedValidator
#GestationalAge
aMeasurement
type
value

aQuantity
value <36>
unit <weeks>
anObserv-Type
<#COMP-GAL>
aCompObs

aRangedValidator
aMeasurement anObser-Type
<aQuantity> <#GAL>

anotherRangedValidator
anotherMeasurement anObser-Type
<anotherQuantity> <#UDT>

Slide - 47

Observations: TypeObject

ObservationType TypeObject
phenomenonType 1 descriptor 1..* Observation
isValid: obsValue instance recordedDate
1..* elements comments
elements 1..* isValid
Quantity
PrimitiveObservation CompositeObservation unit
Type Type value 1..* quantity
CompositeObservation Primitive Observation

Knowledge level DiscreteValues
dvalue 1..*
Operational level

Slide - 48

Observations: Properties
Party

properties
1..*
1..* Observation
instance recordedDate
comments
Quantity
unit
value 1..* quantity
Knowledge level
dvalue 1..* DiscreteValues
Operational level

Slide - 49

Observations: TypeSquare
contDescr 1
ObservationType 1..* varType PartyType
phenomenonType 1 descriptor
isValid: obsValue 1 descriptor 1..*
1..* elements
Party

Type Type properties
1..*
1..* Observation
comments
Quantity
unit
value 1..* quantity
Knowledge level

Operational level dvalue 1..* DiscreteValues

Slide - 50

Observations: Strategy

1..* type
ObservationType Validator
phenomenonType guard 1
validatorName
isValid: obsValue isValid: obsValue
1..* elements

PrimitiveObservation CompositeObservation DiscreteValidator RangedValidator NullValidator
Type Type descriptorSet intervalSet
validUnit

Knowledge level
Operational level

Slide - 51

Medical Observations Design
1..* type contDescr 1
Entities guard 1
Validator
validatorName
ObservationType 1..* varType PartyType isValid: obsValue
phenomenonType 1 descriptor
isValid: obsValue 1 descriptor 1..*
1..* elements
Party DiscreteValidator RangedValidator NullValidator
PrimitiveObservation CompositeObservation validUnit
Type Type properties
1..*
1..* Observation
Behavior
comments
Quantity
unit
value 1..* quantity
Knowledge level

Operational level dvalue 1..* DiscreteValues

Attributes

Slide - 52

Strategies/Interpreters/RuleObjects
(Behavior/Methods)
SomeStrategy Entity
1
*
-sharedAttributes : someType -specificAttribues : type
+sharedInterface() * +someOperations()

RuleObject
Strategy1 Strategy2
... StragegyN
*
1

1

Strategy2.1 Strategy2.2 PrimitiveRule CompositeRule

ANDCondition ORCondition NOTCondition

Design Patterns - GOF95
Composite Strategies  Interpreter Slide - 53

Composite Strategies

Problem: Strategy leads to a big
class hierarchy, one class for
each kind of policy.

Solution: Make Composite
Strategies using Primitive
Operations.

=> Interpreter pattern
Slide - 54

Adaptive Object Model
“Common Architecture”
-children * * -children

Accountability -type
1
AccountabilityType
Classes with
1
Attributes and
1 1 Relationships
1 -accountabilities 1 -accountabilitieTypes
Behavior
-children ** -children * *
-rules *
Entity -type EntityType Rule

1 1 1
1 1 +valueUsing:()

1 1 -attributeTypes

-attributes * *
* 1
*
Attribute AttributeType Constant CompositeRule
+name +value
-type +type 1 * 1
+valueUsing:()

*

Operational Knowledge (meta) TableLookup BinaryOperation

*

Slide - 55

Process for Developing
Adaptable Systems
Developed Iteratively and Incrementally.
 Be Agile, Scrum, XP, Retrospective
Get Customer Feedback early and often.
 User Scenarios – User Stories…
Add flexibility only when and where needed.
Provide Test Cases and Suites for both the
Object-Model and the Meta-Model.
Develop Support Tools and Editors for
manipulating the metadata.
Very similar to Evolving Frameworks
by Roberts and Johnson PLoPD3
Slide - 56

Refactoring Addresses Some
Key Leverage Points
Refactoring is a technique that works with Brooks’
“promising attacks” (from “No Silver Bullet”):
 buy rather than build: restructuring interfaces to support
commercial SW
 grow don’t build software: software growth involves
restructuring (isn’t this core to Agile???)
 requirements refinements and rapid prototyping:
refactoring supports such design exploration, and
adapting to changing customer needs
 support great designers: a tool in a designer’s tool chest.

Slide - 57

Papers and Web Sites
Bill Opdyke’s Thesis
Don Robert’s Thesis
 http://st-www.cs.illinois.edu/users/droberts/thesis.pdf
Refactoring Browser
 http://st-www.cs.illinois.edu/users/brant/Refactory
Evolving Frameworks
 http://st-www.cs.uiuc.edu/users/droberts/evolve.html
Extreme Programming
 http://www.c2.com/cgi-bin/wiki?ExtremeProgramming

Slide - 58

More Web Sites
Wiki wiki web
 http://c2.com/cgi/wiki?WikiPagesAboutRefactoring

The Refactory, Inc.
 http://www.refactory.com

Martin Fowler’s Refactoring Pages
 http://www.refactoring.com/

Adaptive Object Models
 http://www.adaptiveobjectmodel.com/

Slide - 59

That’s All

Slide - 60

Refactoring AOMs For AgilePT2010

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Refactoring AOMs For AgilePT2010 (20)

Refactoring AOMs For AgilePT2010