Analysis Patterns

Martin Fowler

fowler@
[email protected]
acm.org
http://ourworld
http://ourworld..compuserve.com/homepages/Martin_Fowler
compuserve.com/homepages/Martin_Fowler

Page 1 © Martin Fowler 9/11/99

What we will cover

Introduction

Observation
Patterns

Accounting
Patterns

How to use
Patterns

Page 2 © Martin Fowler 9/11/99

1
 Patterns
…projects fail despite the latest technology
for lack of ordinary solutions.

Ralph Johnson and Ward Cunningham

q “Gang of Four”: Design Patterns book

q Pattern Languages of Programming (PLoP)
conferences

A pattern is an idea that has been useful in one practical context

and will probably be useful in others.
[Fowler]

http://www.hillside.net/patterns

Page 3 © Martin Fowler 9/11/99

Design Pattern: Mediator

Mediator Colleague

Concrete Colleague

Concrete Mediator

Concrete Colleague

q Define an object
that encapsulates
how a set of objects
interact.
Gamma, E., Helm, R., Johnson, R. and Vlissides, J. Design
Patterns: elements of reusable object-oriented software,
Addison-Wesley, Reading, MA, 1995.
www.industriallogic.com/papers/learning.html
Page 4 © Martin Fowler 9/11/99

2
 Analysis Patterns
q David Hay
â Relational data models
â Not connected with
patterns community, but
still patterns
â Hay, D. Data Model
Patterns: conventions of
thought, Dorset House,
New York, NY, 1996.
q Martin Fowler
â OO conceptual models
â Healthcare, Accounting,
Financial Trading, Planning
â Fowler, M. Analysis
Patterns: reusable object
models, Addison-Wesley,
Reading MA, 1997.
Page 5 © Martin Fowler 9/11/99

Notation
Each customer Each order
may have has a single
many orders customer
Order
Customer placedOn : Date
1 [
dispatch()

Priority
Order

Individual Corporate
Customer Customer An order
may be a
priority order
A customer may be a
corporate customer or an
individual customer

q Interface not implementation

Martin Fowler with Kendall Scott, UML Distilled: Applying the
Standard Object Modeling Language, Addison-Wesley 1997

Page 6 © Martin Fowler 9/11/99

3
 Patterns in 3 tier architecture
application
1

application
domain database
2

application
3

q Visibility is from application to domain

q Domain carries shared services
q Applications do not see database
q Patterns used in domain tier

Page 7 © Martin Fowler 9/11/99

Observation Patterns

Introduction

Observation
Patterns

Accounting
Patterns

How to use
Patterns

Page 8 © Martin Fowler 9/11/99

4
 Attributes of Person

Person
height : Number
weight : Number
blood glucose level : Number

What are the problems with this model?

Page 9 © Martin Fowler 9/11/99

Quantity
Quantity

amount: Number
Person
units: Unit
height : Quantity
weight : Quantity +, -, *, /
blood glucose level : Quantity <, >, =, <=, >=
as(Unit) : Quantity
toString() : String
valueOf(String) : Quantity

q Define arithmetic operations for quantity

q Money is a kind of quantity
q Use Conversion Ratio to convert to other
units

What if there are tens or hundreds of attributes?

Page 10 © Martin Fowler 9/11/99

5
 Measurement
Phenomenon Type

- Martin Fowler is six feet tall

1

[
Person
[ Measurement

amountOf(PhenomenonType) : Quantity amount: Quantity

1

q For each attribute of person, define an

instance of phenomenon type
q To record a value for a person create a
measurement
When should we use this model?
Page 11 © Martin Fowler 9/11/99

Observation
1
Phenomenon Phenomenon Type
[
1 1

[
Category
Observation
[
Observation
Measurement
{overlapping}
[ amount: Quantity

1
Person

amountOf(PhenomenonType) : Quantity
valueOf(PhenomenonType) : Phenomenon

- Martin Fowler is blood group O

- Martin Fowler is Male

Page 12 © Martin Fowler 9/11/99

6
Knowledge and Operational Levels
arguments
[ [
Associative
Phenomenon
Function
[ 1
result
1
Knowledge

Operational

[
Category
Person Observation
Observation
1 [

q Operational
â regular day to day objects
q Knowledge
â Meta-Data
â Objects that capture domain rules
Page 13 © Martin Fowler 9/11/99

Range
Range Magnitude
upper: Magnitude
lower: Magnitude =
isUpperInclusive: Boolean <
isLowerInclusive: Boolean =<
>
includes (Magnitude) : Boolean >=
overlaps (Range) : Boolean
abuts (Range) : Boolean

- Between 10 miles and 50 miles

- 6m < x ≤ 15m

Page 14 © Martin Fowler 9/11/99

7
 Phenomenon with Range
1
Range Phenomenon

1
Phenomenon
{phenomena's ranges
Type
should not overlap}

- Normal Heart Rate is 60–80 beats per minute

Page 15 © Martin Fowler 9/11/99

Assigning a phenomenon
new heart rate: normal heart rate: fast heart rate:
a Measurement
Phenomenon Type Phenomenon Phenomenon

phenomenon = phenomenonIncluding(aMeasurement)

includes (aMeasurement)

false

includes (aMeasurement)

true

fastHeartRate

setPhenomenon(fastHeartRate)

Page 16 © Martin Fowler 9/11/99

8
 Corporate Finance Problem

q Each month we get a large number of

indicators (revenue, income, units sold…)
q Each indicator is given values for good,
reasonable, poor, or bad
â If revenue is over 105 million that is good,
over 95 is reasonable, over 90 is poor and
under 90 is bad.
q Indicators can be calculated based on
other indicators
â Revenue = income - costs
â Income = units sold * average price

Page 17 © Martin Fowler 9/11/99

Measurement Protocol
[ 1 Measurement [ 1 Phenomenon
Measurement
Protocol {list} Type

Source
Calculated
Measurement
Measurement
Protocol
Protocol

q Source Protocols read from external

systems
q Calculated Protocols are calculations

Page 18 © Martin Fowler 9/11/99

9
 Calculated Measurement
result type
Measurement 1 Phenomenon
Protocol [ Type

1 Calculated argument types {list}

Formula Measurement
[ Protocol [

q Calculated measurement protocol has

formula and arguments
- Revenue = Income - Costs
- Income = Number sold X Average Price
- Body Mass Index = weight / height2

Page 19 © Martin Fowler 9/11/99

Creating a Measurement
a Measurement
a Formula
Protocol
create Measurement

find arguments

execute
return quantity

new
a Measurement

q Factory method for measurements

Page 20 © Martin Fowler 9/11/99

10
 Observation Patterns

q Quantity
q Measurement
q Observation
q Range
q Phenomenon with Range
q Measurement Protocol

Page 21 © Martin Fowler 9/11/99

Accounting Patterns

Introduction

Observation
Patterns

Accounting
Patterns

How to use
Patterns

Page 22 © Martin Fowler 9/11/99

11
 Account
Entry
Account
amount: Money
balance: Money 1 [ whenCharged: Timepoint
whenBooked: Timepoint

{balance = sum(entries)}

q Entries record history of changes to an

account

Where might we use this model?

Page 23 © Martin Fowler 9/11/99

Transaction
Entry
1 [ amount: Money 2..[ 1
Account Transaction

{sum(entries.amount) = 0}

- I withdraw $100 from my checking account

- The Willenhall Coal Company delivers 30 tons of coal to Bilston
Gasworks

Why are transactions useful?

Page 24 © Martin Fowler 9/11/99

12
 Posting Rule (simple)

Posting Rule
1
Account multiplier: Number
output
execute()
1
trigger

- 5% of state tax is deducted from gross income

Page 25 © Martin Fowler 9/11/99

Individual Instance Method

Posting Rule
1 Method
calculation

- Income tax is deducted at

15% for first $31,500
output trigger 28% to $80,750
1 1 31% to $130,800

Account …

q Each instance of posting rule has its own

method
q Conceptual statement

Page 26 © Martin Fowler 9/11/99

13
 Account Types
output 1
Account Type Posting Rule Method
[
1 1
trigger
Knowledge

Operational

[
Employee
1
Employee
Account [

- Apply the graduated tax algorithm to the gross pay of Doug and
Dinsdale Piranha to yield their net pay.

Page 27 © Martin Fowler 9/11/99

Network of Posting Rules

raw Overtime:
3 hours (Tuesday)
Account Type

straight += input
OT hours PR if (Sunday) excess += input
else excess += input * 0.5

straight OT hours : Excess Hours:

3 hours 1.5 hours
Account Type Account Type

excess = input *
employee.otRate * 2/3
premium = input *
employee.otRate * 1/3
Straight OT Payment out = input * Excess Payment:
: Posting Rule employee.rate Posting Rule

Straight Dollars : Excess Hours: Premium Hours:

$150 $50 $25
Account Type Account Type Account Type

Total Overtime :
$225 Summary Account
Type

Page 28 © Martin Fowler 9/11/99

14
 Executing the network
a list of Posting Rules a Posting Rule an Employee input Account output Account

pay (employee) *[for all elements]

run (employee) get account (input account type)

get unprocessed entries

* [for each entry]

execute posting rule

post entry

Page 29 © Martin Fowler 9/11/99

Procedural Approach
H
H

Get straight hours

account
Process Hours

get unprocessed
Dollarize Straight entries
Hours

[for all entries]

[

multiply by rate
Dollarize Excess
Hours

H post to straight
dollars

Page 30 © Martin Fowler 9/11/99

15
 Changing the Rules
H
H

Get straight hours

Process Hours account

get unprocessed
Dollarize Straight entries
Hours

[ [for all entries]

[on or after April 15]

Dollarize Excess
[before April 15]
Hours

multiply by rate multiply by rate * 0.9

post to straight
dollars

Page 31 © Martin Fowler 9/11/99

Adding an effectivity time

Posting Rule
output 1
Account Type effectivity: Date Range Method
[
1 1
trigger

[
Employee
1
Employee
Account [

q Add a date range for each posting rule

instance

Page 32 © Martin Fowler 9/11/99

16
 New Behavior
a list of Posting Rules a Posting Rule an Employee input Account output Account

pay (employee) *[for all elements]

run (employee) get account (input account type)

get unprocessed entries in (date range)

* [for each entry]

execute posting rule

post entry

Page 33 © Martin Fowler 9/11/99

Implementing with singleton

State Tax PR
state: State
calculate()

{Singleton} {Singleton} {Singleton}

CA Tax PR MA Tax PR NY Tax PR
state: State state: State state: State
calculate() calculate() calculate()

What are the strengths and weaknesses?

Page 34 © Martin Fowler 9/11/99

17
 Strategy Pattern
{abstract}
State Tax PR Calculation Strategy
1
state: State
calculate() calculate()

«Method»
return strategy.calculate()

{Singleton} {Singleton} {Singleton}

CA Tax MA Tax NY Tax

calculate() calculate() calculate()

What are the strengths and weaknesses?

Page 35 © Martin Fowler 9/11/99

Parameterized Method
State Tax PR
state: State
rate: Number
calculate() «Method»
multiply by rate

What are the strengths and weaknesses?

Page 36 © Martin Fowler 9/11/99

18
 Combining Implementations
State Tax PR
state: State
calculate()

{Singleton} {Singleton}
CA Tax PR Multiplier PR NYTax PR
rate: Number

calculate() calculate() calculate()

q Combination of singleton and

parameterized approaches

Page 37 © Martin Fowler 9/11/99

Posting Rule Execution

q Eager firing
â Fire when entry added to account
â Find all posting rules looking at that account
and fire them (etc, etc)
q Posting rule based
â Have a coordinator which tells posting rules
when to fire
q Account based based
â Have a coordinator tell which accounts when
to fire their outbound posting rules
q Backward Chained
â When you ask an account for its balance
â All input accounts must be asked for value
(etc, etc)

Which would you choose?

Page 38 © Martin Fowler 9/11/99

19
 Accounting Patterns

q Account
q Transaction
q Posting Rule
q Individual Instance Method
q Posting Rule Execution
q Posting Rules for Many Accounts
q Accounting Practice

Page 39 © Martin Fowler 9/11/99

How to use Patterns

Introduction

Observation
Patterns

Accounting
Patterns

How to use
Patterns

Page 40 © Martin Fowler 9/11/99

20
 Using Patterns
q Starting point for development
â Inspiration to get things going
q Comparison for review
â Why are we different to this pattern?
q Documentation of complex frameworks
â Highlight the key patterns
q Informal standardization
q Mining from legacy systems
q Training
â The next step after basic training

The why is at least as important as the what

Page 41 © Martin Fowler 9/11/99

Learning about Patterns

q Read the books
â Use a book group
â Discuss two chapters a week over lunch
q Take a training course
â Training must emphasize using patterns
â Instructors expertise is central
q Try using the patterns
â You cannot understand a pattern till you
have tried it
â Trying a pattern and finding it does not fit is
valuable
q Write some patterns

http://www.hillside.net/patterns

Page 42 © Martin Fowler 9/11/99

21
 Books
q Beck, K. Smalltalk Best Practice Patterns. Volume 1: Coding,
Prentice Hall, Englewood Cliffs, NJ, 1997.
q Buschmann, F., Meunier, R., Rohnert, H., Sommerlad, P. and
Stal., M. Pattern-Oriented Software Architecture - A System of
Patterns, John Wiley, 1996.
q Coad, P., North, D. and Mayfield, M. Object Models: strategies,
patterns and applications, Prentice Hall, Englewood Cliffs,
1995.
q Coplien, J.O. and Schmidt, D.C. Pattern Languages of Program
Design, Addison-Wesley, Reading, MA, 1995.
q Gamma, E., Helm, R., Johnson, R. and Vlissides, J. Design
Patterns: elements of reusable object-oriented software,
Addison-Wesley, Reading, MA, 1995.
q Fowler, M. Analysis Patterns, Addison-Wesley, Reading MA,
1997
q Hay, D. Data Model Patterns: conventions of thought, Dorset
House, New York, NY, 1996.
q Vlissides, J.M., Coplien, J.O. and Kerth, N.L. ed.Pattern
Languages of Program Design 2, Addison-Wesley, 1996.

http://www.hillside.net/patterns

Page 43 © Martin Fowler 9/11/99

Final Thoughts
q Patterns are relevant for all aspects of
software engineering
â and for Business Process Reengineering
q Analysis patterns can be used across
traditional vertical business domains
q A model is not right or wrong, only more
or less useful
q Patterns are a starting point, not a final
destination
q Still learning about teaching, using, and
discovering patterns

Page 44 © Martin Fowler 9/11/99

22