Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises

2004 Prentice Hall November 14, 2003

1. Introduction to Software Engineering: Solutions

1-1 What is the purpose of modeling?

The purpose of modeling is to reduce complexity by building a simplied representation of reality which ignores
irrelevant details. What is relevant or not is dened by the questions the model will be used to answer.

12 A programming language is a notation for representing algorithms and data structures. List two advantages
and two disadvantages of using a programming language as sole notation throughout the development process.

Advantages:
Developers need only learn one notation for all development activities.
Traceability among models and between models and code is made easier since they are written in the same
notation.
Disadvantages:
A programming language is a low level notation which is difcult to use for representing user requirements, for
example.
A programming language enables and encourages developers to represent implementation details too early.

13 Consider a task you are not familiar with, such as designing a zero-emissions car. How would you attack the
problem?

This is an open ended question whose purpose is for students think about problems they cannot solve without help.
Answers should contain two or more of the following points:
Dene the problem precisely by gathering information from potential users.
Discover the boundaries of the solution space by gathering information from application domain experts.
Brainstorm ideas with other people, including experts and non experts
Evaluate ideas using prototypes, simulations, and candidate users.

14 What is meant by knowledge acquisition is not sequential? Provide a concrete example of knowledge
acquisition that illustrates this.

Knowledge acquisition is nonlinear in the sense that the acquisition of a new piece of knowledge may invalidate prior
knowledge. In other terms, knowing one more piece of information may lead you to realize that what you thought you
knew is invalid. Galileo Galilei invalidated the earth centric model of the universe by observing the moons of Jupiter
and the phases of Venus.

15 Hypothesize a rationale for the following design decisions:

This exercise tests if the student understands the difference between a decision and its rationale. The exact rationale
provided by the student is not important as long as it is rationale (e.g., the answer to the rst bullet could have been to
allow snow whites seven dwarves to purchase train tickets).

The ticket distributor will be at most one and a half meters tall.

Enable children and persons in wheelchair to purchase tickets.

The ticket distributor will include two redundant computer systems.

To achieve a high level of availability such that ticket distribution is not interrupted (and thus, ticket sales not lost in
the case of the failure of one computer).

The ticket distributor will include a touch screen for displaying instructions and inputing commands.
The only other control will be a cancel button for aborting a transaction.

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Enable substantial modications to the interface (e.g., increase the number of tariff zones or the number of products)
without changes to the hardware.

16 Specify which of the following statements are functional requirements and which are nonfunctional
requirements:
The ticket distributor must enable a traveler to buy weekly passes.
The ticket distributor must be written in Java.
The ticket distributor must be easy to use.

The rst requirement is functional, the third is nonfunctional. Using the denitions in Chapter 1, the second
requirement is nonfunctional. In Chapter 4, we will call this requirement a pseudo requirement as it constrains aspects
of the system that are not visible to the user.

17 Specify which of the following decisions were made during requirements or system design:
The ticket distributor is composed of a user interface subsystem, a subsystem for computing tariff, and a
network subsystem managing communication with the central computer.
The ticket distributor will use PowerPC processor chips.
The ticket distributor provides the traveler with an on-line help.

The rst decision is a system design decision. The second decision is also a system design decision if made by
developers (otherwise, it is a requirements decision). The third decision is a requirements decision.

18 In the following description, explain when the term account is used as an application domain concept and when
as a solution domain concept:
Assume you are developing an online system for managing bank accounts for mobile customers. A major
design issue is how to provide access to the accounts when the customer cannot establish an online connection.
One proposal is that accounts are made available on the mobile computer, even if the server is not up. In this
case, the accounts show the amounts from the last connected session.

The rst two occurrences of account are application domain concepts while the last two occurrences are solution
domain concepts. The phrases accounts are made available on the mobile computer and the accounts show the
amounts from the last connected session denote a solution domain concept that gives users the illusion that they are
accessing their bank accounts on their mobile computer. However, the actual bank account is not on the mobile
computer.

19 What is the difference between a task and an activity?

An activity is composed of a number of tasks. Both represent work, but tasks cannot conveniently be decomposed any
further.

110 A passenger aircraft is composed of several millions of individual parts and requires thousands of persons to
assemble. A four-lane highway bridge is another example of complexity. The rst version of Word for Windows,
a word processor released by Microsoft in November 1989, required 55 person-years, resulted into 249,000
lines of source code, and was delivered 4 years late. Aircraft and highway bridges are usually delivered on time
and below budget, whereas software is often not. Discuss what are, in your opinion, the differences between
developing an aircraft, a bridge, and a word processor, which would cause this situation.

This is an open question whose purpose is to have students realize that software systems are not the only complex
systems out there. Answers should include two or more of the following points:
To estimate the budget and schedule for a new bridge or aircraft, engineers use actual data from previous bridges
and aircraft. Word for Windows was an innovative piece of software with few or no precedents.
Many bridges and aircraft are simply renements of other existing artifacts. This reduces the proportion of the
overall effort that is dedicated to design (which is the most difcult to estimate).
Bridges and aircraft are often associated with severe nancial penalties when late or over budget.

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Bridges and aircraft have safety requirements associated with them. This leads to a conservative approach to
development including the use of mature technologies and well dened processes.
Bridges and aircraft are sometimes delivered late too.

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003

2. Modeling with UML: Solutions

2-1 Consider an ATM system. Identify at least three different actors that interact with this system.

An actor is any entity (user or system) that interacts with the system of interest. For an ATM, this includes:
Bank Customer
ATM Maintainer
Central Bank Computer
Thief
The last actor is often referred to as a misactor in the literature, because it is an actor that interacts with the system
but shouldnt.

22 Can the system under consideration be represented as an actor? Justify your answer.

The system under consideration is not external to the system and shouldnt be represented as an actor. There are a few
cases, however, when representing the system as an actor may clarify the use case model. These include situations
where the system initiates uses cases, for example, as time passes (Check for Outdated Articles, Send Daily
Newsletter).

23 What is the difference between a scenario and a use case? When do you use each construct?

A scenario is an actual sequence of interactions (i.e., an instance) describing one specic situation; a use case is a
general sequence of interactions (i.e., a class) describing all possible scenarios associated with a situation. Scenarios
are used as examples and for clarifying details with the client. Use cases are used as complete descriptions to specify
a user task or a set of related system features.

24 Draw a use case diagram for a ticket distributor for a train system. The system includes two actors: a traveler,
who purchases different types of tickets, and a central computer system, which maintains a reference database
for the tariff. Use cases should include:

BuyOneWayTi cket

,

BuyWeekl yCar d

,

BuyMont hl yCar d

,

Updat eTar i f f

.
Also include the following exceptional cases:

Ti me- Out

(i.e., traveler took too long to insert the right amount),

Tr ansact i onAbor t ed

(i.e., traveler selected the cancel button without completing the transaction),

Di st r i but or Out Of Change

, and

Di st r i but or Out Of Paper

.

Figure 2-1 Example solution to Exercise 24.
Traveler
CentralComputerSystem
BuyOneWayTicket
BuyWeeklyCard
BuyMonthlyCard
UpdateTariff
Time-Out
TransactionAborted
DistributorOutOfChange
DistributorOutOfPaper
DistributorException
<<extend>>
<<extend>>
<<extend>>

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
This questions can have several correct answers, Figure 2-1 being a possible answer. The following elements should
be present:
The relationship between an actor and a use case is a communication relationship (undirected solid line).
The relationship between exceptional use cases and common use cases is an

<<extend>>

relationship.
The exceptional use cases described in the exercise only apply to the use cases invoked by the traveler.
The following elements should be present in a good answer:
All exceptions apply to all traveler use cases. Instead of drawing 3x4 relationships between these use cases, an
abstract use case from which the exceptional use case inherit can be used, thus reducing the number of

<<extend>>

relationships to 3 at the cost of introducing 4 generalization relationships.
The student can introduce exceptional use cases not specied in the exercise that apply to the

CentralComputerSystem

use cases.

25 Write the ow of events and specify all elds for the use case UpdateTariff that you drew in Exercise 24. Do
not forget to specify any relationships.

Figure 2-2 depicts a possible solution for this exercise.

Use case name

UpdateTariff

Participating actor

Initiated by

CentralComputerSystem

Flow of events

1. The

CentralComputerSystem

activates the UpdateTariff function of the ticket distributors available
on the network.
2. The ticket distributor disables the traveler interface and posts a sign indicating
that the ticket distributor is under maintenance.
3. The ticket distributor waits for the new database from the

CentralComputerSystem

.
4. After waiting a minute for the ticket distributors to reach a waiting state, the

CentralComputerSystem


broadcasts the new database.
5. The ticket distributor system receives the new database of tariff. Upon complete,
the ticket distributor sends an acknowledgement to the

CentralComputerSystem

.
6. After acknowledgment, the ticket distributor enables the traveler interface and
can issue tickets at the new tariff.
7. The

CentralComputerSystem

checks if all ticket distributors have acknowledged the new database. If
not, the CentralComputerSystem invokes the

CheckNonRespondingDistributors

use case.

Entry condition

The ticket distributor is connected to a network reachable by the CentralComputerSystem.

Exit condition

The ticket distributor can issue tickets under the new tariff, OR
The ticket distributor is disabled and displays a sign denoting that it is under maintenance.

Quality
requirements

The ticket distributor stays ofine at most 2 minutes and is considered out-of-order otherwise.

Figure 2-2 A possible solution for the

UpdateTariff

use case.

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003

26 Draw a class diagram representing a book dened by the following statement: A book is composed of a
number of parts, which in turn are composed of a number of chapters. Chapters are composed of sections.
Focus only on classes and relationships.

This exercise checks the students understanding of basic aspects of class diagrams, including:
Classes are represented with rectangles.
The attribute and operations compartment can be omitted.
Aggregation relationships are represented with diamonds.
Class names start with a capital letter and are singular.

27 Add multiplicity to the class diagram you produced in Exercise 26.

See Figure 2-3. Aggregation does not imply multiplicity, thus the

1..*

multiplicity is necessary.

28 Draw an object diagram representing the rst part of this book (i.e., Part I, Getting Started). Make sure that the
object diagram you draw is consistent with the class diagram of Exercise 26.

This exercise checks the students understanding of basic aspects of object diagrams, including:

Figure 2-3 Example solution for Exercise 26.

Figure 2-4 Example solution for Exercise 28.
Book
Part
Chapter
Section
1
*
1
*
1
*
gettingStarted:Part
1.1:Section
1.2:Section
1.3:Section
1.4:Section
1.5:Section
1.6:Section
refs:Section
2.1:Section
2.2:Section
2.3:Section
2.4:Section
2.5:Section
refs:Section
3.1:Section
3.2:Section
3.3:Section
3.4:Section
3.5:Section
3.6:Section
refs:Section
oose:Book
uml:Chapter introduction:Chapter communication:Chapter

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Objects are represented with rectangles and underlined labels.
The class of an object is included in the label of the object (e.g.,

uml:Chapter

is of class

Chapter

).
Links are represented with solid lines

29 Extend the class diagram of Exercise 26 to include the following attributes:
a book includes a publisher, publication date, and an ISBN
a part includes a title and a number
a chapter includes a title, a number, and an abstract
a section includes a title and a number

This exercises checks the students knowledge of attributes and their representation in UML (page 45).

210 Consider the class diagram of Exercise 29. Note that the

Part

,

Chapter

, and

Section

classes all include a title
and a number attribute. Add an abstract class and a generalization relationship to factor out these two
attributes into the abstract class.

This exercise checks the students knowledge of abstract classes and inheritance.

211 Draw a class diagram representing the relationship between parents and children. Take into account that a
person can have both a parent and a child. Annotate associations with roles and multiplicities.

Figure 2-7 depicts a canonical solution. This exercise is meant to emphasize the difference between a relationship, a
role, and a class. In the above sentence, parent and child are roles while person is the class under consideration. This
results in a class diagram with a single class and a single association with both ends to the class.

Figure 2-5 Example solution for Exercise 29.

Figure 2-6 Example solution for Exercise 210.
Book
publisher:Person
publicationDate:Date
ISBN:Integer[4]
Part
title:String
number:Integer
Chapter
title:String
number:Integer
abstract:String
Section
title:String
number:Integer
1 * 1 * 1 *
Book
publisher:Person
publicationDate:Date
ISBN:Integer[4]
Part Chapter
abstract:String
Section
1 * 1 * 1 *
Number edComponent
title:String
number:Integer
abstract:String

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
A common modeling mistake by novices is to draw two classes, one for the parent and one for the child.

212 Draw a class diagram for bibliographic references. Use the references in Appendix C, Bibliography, to test
your class diagram. Your class diagram should be as detailed as possible.

The domain of bibliographic references is rich and complex. Consequently, students should deepened their
understanding of the domain before they attempt to draw a class diagram (similar to the requirements analysis of a
system). Figure 2-8 depicts an incomplete sample solution that could be accepted as sufcient from an instructor. The
class diagram should minimally include the following concepts:
An abstract class

Publ i cat i on

(or

Bi bl i ogr aphi cRef er ence

)
A many to many relationship between

Author

and

Publ i cat i on

At least three or more concrete classes rening

Publ i cat i on

At least one aggregation relationship (e.g., between

Journal

and

Article

or

Proceedings

and

ConferencePaper

). Both ends of the aggregation should also be subclasses of

Publ i cat i on

.

213 Draw a sequence diagram for the

warehouseOnFire

scenario of Figure 2-15. Include the objects

bob

,

al i ce

,

j ohn

,

FRI END

, and instances of other classes you may need. Draw only the rst ve message sends.

Figure 2-7 Example solution for Exercise 211.

Figure 2-8 Example solution for Exercise 212.
Person
parent
child
*
2
Book
publisher:Person
publicationDate:Date
ISBN:Integer[4]
Journal
volume:Integer
Proceedings
Publ i cat i on
title:String
publicationDate:Date
Article
firstPage:Integer
CompendiumChapter
chapterNumber
ConferencePaper
number:Integer
1
*
1
*
Compendium
editors:Person[]
1
*
lastPage:Integer
conferenceLocation
conferenceDate:Date
firstPage:Integer
lastPage:Integer
Author
* *

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
This exercise checks the students knowledge of sequence diagrams when using instances. This exercise can have
several correct answers. In addition to the UML rules on sequence diagrams, all correct sequence diagrams for this
exercise should include one or more actors on the left of the diagram who initiate the scenario, one or more objects in
the center of the diagram which represent the system, and a dispatcher actor on the right of the diagram who is
notied of the emergency. All actors and objects should be instances. Figure 2-9 depicts a possible answer.

214 Draw a sequence diagram for the

Repor t I nci dent

use case of Figure 2-14. Draw only the rst ve message
sends. Make sure it is consistent with the sequence diagram of Exercise 213.

This exercise tests the students knowledge of sequence diagram when using classes. Like the previous exercise, this
exercise can have several correct solutions. A correct solution should include a FieldOfcer actor on the left, a
Dispatcher actor on the right, and one or more classes in the middle representing the FRIEND system. The answer to
this exercise should be consistent with the answer to the previous exercise, in the sense that all class and operation
names should be the same. Figure 2-10 depicts a possible answer.

215 Consider the process of ordering a pizza over the phone. Draw an activity diagram representing each step of
the process, from the moment you pick up the phone to the point where you start eating the pizza. Do not
represent any exceptions. Include activities that others need to perform.

Figure 2-11 is an example solution for this exercise. The following elements should be present in the solution:
Activity names should be verb phrases indicating what the initiating actor is attempting to accomplish. Roles
should be indicated with swimlanes

Figure 2-9 Example solution for Exercise 213.

Figure 2-10 Example solution for Exercise 214.
:FRIEND
:EmergencyForm
reportEmergency()
new EmergencyForm()
specifyIncident()
notifyDispatcher()
alice john
requestResource(fireTruck)
commit()
bob
FRIEND
EmergencyForm
reportEmergency()
new EmergencyForm()
specifyIncident()
notifyDispatcher()
FieldOfficer Dispatcher
*requestResource()
commit()

Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Activities that are concurrent or which do not need to happen in a sequential order should be indicated with
complex transitions.

216 Add exception handling to the activity diagram you developed in Exercise 215. Consider at least three
exceptions (e.g., delivery person wrote down wrong address, deliver person brings wrong pizza, store out of
anchovies).

This exercise checks the students knowledge of decision points. Figure 2-12 depicts an example solution. Students
modeling a realistic example will also think about cascaded exceptions and learn to represent them with multiple
decision points.
217 Consider the software development activities which we described in Section 1.4. Draw an activity diagram
depicting these activities, assuming they are executed strictly sequentially. Draw a second activity diagram
depicting the same activities occurring incrementally (i.e., one part of the system is analyzed, designed,
implemented, and tested completely before the next part of the system is developed). Draw a third activity
diagram depicting the same activities occurring concurrently.
This exercise tests the students knowledge of the activity diagram syntax, not the knowledge of software engineering
activities. In the sample solution provided in Figure 2-13, we assumed that requirements elicitation need to be
completed before any subsystem decomposition can be done. This leads to a splitting of the ow of control in the
third diagram where all remaining activities occur concurrently. The instructor could accept a solution where
requirements elicitation also occurs incrementally or concurrently with the other development activities.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Figure 2-11 An example of pizza order activity diagram with exception handling.
Call
Pizza Shop
Ask for
pizza size
Client
Ask for
toppings
Ask for
pizza type
Ask for
address&phone
Wait for
pizza
Prepare
pizza
Deliver
pizza
Receive
pizza
Pay
Eat
Pizza Shop
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Figure 2-12 An example of pizza order activity diagram.
Call
Pizza Shop
Ask for
pizza size
Client
Ask for
toppings
Ask for
pizza type
Ask for
address&phone
Wait for
pizza
Prepare
pizza
Deliver
pizza
Receive
pizza
Pay
Eat
Pizza Shop
Ask correct
address
Take
loss
[wrong address]
[wrong address]
Deliver to
new address
[wrong pizza]
[no anchovies]
Revise
topping
[cancel order]
[proceed with order]
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Figure 2-13 Example solution for Exercise 217.
Requirements
Elicitation
System
Design
Object
Design
Analysis Implementation
Requirements
Elicitation
System
Design 1
Object
Design 1
Analysis 1
System
Design 2
Object
Design 2
Analysis 2
Implementation
1
Implementation
2
System
Design 3
Object
Design 3
Analysis 3
Implementation
3
Requirements
Elicitation
System
Design 1
Object
Design 1
Analysis 1
System
Design 2
Object
Design 2
Analysis 2
Implementation
1
Implementation
2
System
Design 3
Object
Design 3
Analysis 3
Implementation
3
Sequential activities
Incremental activities
Concurrent activities
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
3. Project Communication: Solutions
3-1 What is the difference between a role and a participant?
A participant is an individual. A role corresponds to a set of responsibilities. A single participant can assume many
roles. Some roles may be shared among more than one participant.
32 Can a role be shared between two or more participants? Why or why not?
Some roles are assigned to more than one participant. For example, a team of developers can be assigned to the same
subsystem. Other roles, such as chief architect or project manager, can only be assigned to a single participant in the
project. In the end, the issue is one of assigning clear and non-overlapping tasks to single participants: When two or
more participants are responsible for exactly the same task, none of them will feel personally responsible for the
success of the task.
33 What is the difference between a client and an end user?
The client contracts the development of a system. The client is responsible for dening the scope of the system and
for nancing the project. The end users are the participants who actually uses the system to accomplish their work. In
some project, the role of end user and client may be shared by the same participant. In general, however, this is not the
case.
34 To which roles would you assign the following tasks?
Change a subsystem interface to accommodate a new requirement. [system architect]
Communicate the subsystem interface change to other teams. [API engineer (liaison)]
Change the documentation as a result of the interface change. [editor]
Design a test suite to nd defects introduced by the change. [tester]
Ensure that the change is completed on schedule. [project manager or team leader]
35 You are responsible for coordinating the development of a system for processing credit applications for a bank.
In what roles would the following project participants be able to contribute most to the project?
a bank employee responsible for processing credit applications [end user]
the manager of the information technology group at the bank, who contracted the system [client]
a free-lancer who developed similar systems in the past [developer]
a technical writer [editor]
36 Draw a UML activity diagram representing the meeting process described in Section 3.4.3. Focus in particular
on the work products generated before and after the meeting, such as the agenda and the meeting minutes. Use
swimlanes to represent roles.
Figure 3-1 depicts an example solution. This exercise tests the students knowledge of the meeting process and of
activity diagrams described in Chapter 2.
37 What is the difference between a work package and a work product?
A work package is a description of work. A work product is the actual result of work (e.g., a document, a piece
of code, etc.).
When is a work package dened?
During planning.
Consider an assignment where two students collaborate to plan and develop a system for sorting lists of names
using two different sort algorithms. The deliverables for the assignment are the source code, the system
documentation, and a manual for other developers explaining how new sorting algorithms can be integrated
into the code. Give examples of work packages and work products in this project.
Examples of work packages:
Plan development
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Dene and document interface between system and sort algorithms
Implement algorithms
Document system
Examples of work products:
development plan
source code
system documentation
manual for explaining how to integrate new sorting algorithms
38 What is the difference between a cross-functional team and a subsystem team? Provide examples and justify
your choices.
Figure 3-1 An example meeting process (UML activity diagram).
Post Initial
Agenda
Facilitator
Review
Agenda
Meeting Participant Minute Taker
Revise
Agenda
Facilitate
Meeting
Take Meeting
Notes
Agree on
Agenda
Review Action
Items
Report Status
Items
Discuss &
Resolve Issues
End
Meeting
Post
Minutes
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
A subsystem team is completely responsible for the development of a subsystem.A cross-functional team is made out
of liaison of subsystem teams and is responsible for support tasks, such as documentation and conguration
management. Also, the system architecture can is often dened by a cross-functional team made out of liaison from
all subsystem teams.
39 As many critical communication events are planned (e.g., client reviews, project reviews, peer reviews), why is
there still a need for unplanned communication events (e.g., request for clarication, request for change, issue
resolution)?
There are many reasons for unplanned communication. Some of the more important ones are:
Reduce latency: While developers could wait for the next project review to request clarication on their tasks or
on interfaces provided by other teams, this would slow down development signicantly and introduce much
rework (e.g., correcting code that was developed under incorrect assumptions).
React to unexpected events: Any system of a realistic size cannot be completely specied and understood a
priori. This results in requirements and design changes during development, which often have to be
communicated and negotiated among several stakeholders. While change processes can be set in place to manage
this communication, much unplanned and informal communication complements the formal communication
events to motivate, clarify, understand, gauge, and negotiate the underlying issues introduced by the change.
Team building: Formal communication events are rarely conducive to a team atmosphere. It is during unplanned
and informal events that participants get know each other and form a community working towards a common
goal.
310 Select a random working day in your work week. Log all activities that qualify as communication activities
(e.g., talking to friends over coffee, obtaining information from a fellow student, providing information,
negotiating, advertising, browsing the web). Which fraction of your working day does communication
represent?
Typically, the fraction of the day spent of communication can easily reach 50% or more. However, there will be a
wide variance in the numbers reported by the students, depending on whether or not they are currently doing project
work.
311 You are a member of the user interface team. You are responsible for designing and implementing forms
collecting information about users of the system (e.g., rst name, last name, address, E-mail address, level of
expertise). The information you are collecting is stored in the database and used by the reporting subsystem.
You are not sure which elds are required information and which are optional. How do you nd out?
This is an open-ended question whose purpose is to have the student think of all the communication paths available.
The side effect of this question is that the student will also have to think about the information sources associated with
a project. A correct answer should contain at least two of the following points:
The user interface team member should rst check the requirements analysis document, described in Chapter 1.
If the answer is not in the requirements analysis document, or if the answer is not clear, the document is
incomplete or does not exist yet, in which case the user interface team member should ask the responsible analyst
on the project. This is a request for change or a request for clarication.
If the analyst does not have the answer, he or she should investigate with a potential user and possibly get
approval from the client. This is also a request for clarication.
In all cases, the answer should eventually make its way into the requirements analysis document.
312 You have been reassigned from the user interface team to the database team due to staff shortages and
replanning. The implementation phase is well underway. In which role would you be most procient given your
knowledge of the user interface design and implementation?
Liaison of the database team to the user interface team.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
313 Assume the development environment is Unix workstations, and the documentation team uses the Macintosh
platform for writing documentation. The client requires the documents to be available on Windows platforms.
Developers produce the design documentation using FrameMaker. The documentation team uses Microsoft
Word for the user-level documentation. The client submits corrections on hardcopies and does not need to
modify the delivered documents. How could the information ow between the developers, the technical writers,
and the client be set up (e.g., format, tools, etc.) such that duplication of les is minimized while everybodys
tool preferences and platform requirements are still satised?
This is a simple real life question with complicated answers. Several approaches are possible, including:
The project uses a printable format as a standard interchange format, such as Postscript or PDF (Portable Display
Format). This allows every participant to access and view documents. Only the author of a document is able to
make changes, however. This makes it difcult to move the document responsibility to move from one team to
another (e.g., from a subsystem team to the documentation team).
The project uses an interchange format such as RTF or XML for all documents. Both formats are recognized by
recent versions of both Microsoft Word and FrameMaker. This allows document responsibilities to be transferred
between developers and technical writers. Unfortunately, this is currently not a practical solution as both tools
understanding of the interchange formats is different and imperfect, resulting in some loss of formatting.
314 Which changes in the organization and communication infrastructure would you recommend for a successor of
the Ariane 5 project as a consequence of the Ariane 501 failure described at the beginning of this chapter?
This is an open-ended question as many solutions exist and detailed knowledge of the current team organization are
not available. Note that this question is only interested in communication or organizational changes, not technical or
methodological changes. A correct answer could include the following points:
Increase communication between the team responsible for system testing and the software developers. This can
be done by introducing liaisons on both teams.
Increase communication between the team responsible for unit testing and for system testing.
Change documentation requirements for reused components.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
4. Requirements Elicitation: Solutions
4-1 Consider your watch as a system and set the time 2 minutes ahead. Write down each interaction between you
and your watch as a scenario. Record all interactions, including any feedback the watch provides you.
This scenario could be written using any watch. Below is an example of a watch with a digital display and two
buttons. Any solution should be detailed enough that an ignorant user can execute the scenario with somebody elses
watch.
42 Consider the scenario you wrote in Exercise 4-1 Identify the actor of the scenario. Next, write the
corresponding use case SetTime. Include all cases, and include setting the time forward, backward, setting
hours, minutes, and seconds.
The following use case is generalized from the scenario of Figure 4-1. The level of detail of the use case should be the
same as the level of detail of the scenario in the previous exercise.
Scenario name
setWatch2MinutesAhead
Participating actor instances
allen:WatchOwner
Flow of events 1. The WatchOwner presses both Watch buttons simultaneously.
2. The Watch enters set time mode and indicates this by blinking the hour digits.
3. The WatchOwner presses the left button once.
4. The Watch stops blinking the hour digits and starts blinking the minutes.
5. The WatchOwner presses the right button twice.
6. The Watch increments the minutes by two.
7. The WatchOwner presses both buttons simultaneously.
8. The Watch stops blinking.
Figure 4-1 The setWatch2MinutesAhead scenario.
Use case name
SetTime
Participating actor Initiated by WatchOwner
Entry condition 1. The Watch is in read time mode.
Flow of events 2. The WatchOwner presses both Watch buttons simultaneously.
3. The Watch enters set hour mode and indicates this by blinking the hour digits.
4. In the set hour mode, the WatchOwner can advance the hour digits by pressing the
right button. Each time the right button is pressed, the hours are incremented by one.
If the hours reach 23 and the right button is pressed, the hours are reset to 0. The date
is not changed.
5. At any time in the set hour mode, the WatchOwner can switch to the set minutes
mode by selecting the left button. To indicate this, the Watch stops blinking the hour
digits and starts blinking the minute digits.
6. In the set minutes mode, the WatchOwner can advance the minute digits by pressing
the right button. Each time the right button is pressed, the minutes are incremented by
one. If the minutes reach 59 and the right button is pressed, the minutes are reset to 0.
The hours are not changed.
Figure 4-2 SetTime use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
43 Assume the watch system you described in Exercises 4-1 and 42 also supports an alarm feature. Describe
setting the alarm time as a self-contained use case named Set Al ar mTi me.
The difference between setting the time of a watch and setting the alarm time should be minimal. A typical answer
can be produced by copying and pasting solution of Exercise 42 and modifying the beginning and the end of the use
case. The rationale behind this similarity is that any user interface should provide similar interfaces for addressing
similar functionality (i.e., user interface consistency). Below is an example generated from the use case of Figure 4-2.
7. At any time in the set minutes mode, the WatchOwner can switch to the set
seconds mode by selecting the left button. To indicate this, the Watch stops blinking
the minute digits and starts blinking the second digits.
8. In the set seconds mode, the WatchOwner can advance the second digits by pressing
the right button. Each time the right button is pressed, the seconds are incremented by
one. If the seconds reach 59 and the right button is pressed, the seconds are rest to 0.
The hours and the minute digits are not changed.
9. At any time in the set seconds mode, the WatchOwner can switch to the set hours
mode by pressing the left button. To indicate this, the Watch stops blinking the second
digits and starts blinking the hour digits.
10. At any time in the set hours, set minutes, and set seconds mode, the WatchOwner
can return to read time mode by pressing both Watch buttons simultaneously.
Exit condition 11. The Watch is in read time mode with the new time set.
Special requirements None.
Use case name
SetAlarmTime
Participating actor Initiated by WatchOwner
Entry condition 1. The Watch is in read time mode.
Flow of events 2. The WatchOwner presses both Watch buttons simultaneously. The Watch enters the set
hour mode. The WatchOwner continues to press both buttons. After one second, a
small alarm bell symbol appears on the display and blinks, indicating that the hours
of the alarm are being set (as opposed to the normal time).
3. In the set hour mode, the WatchOwner can advance the hour digits by pressing the
right button. Each time the right button is pressed, the hours are incremented by one.
If the hours reach 23 and the right button is pressed, the hours are reset to 0. The date
is not changed.
4. At any time in the set hour mode, the WatchOwner can switch to the set minutes
mode by selecting the left button. To indicate this, the Watch stops blinking the hour
digits and starts blinking the minute digits.
5. In the set minutes mode, the WatchOwner can advance the minute digits by pressing
the right button. Each time the right button is pressed, the minutes are incremented by
one. If the minutes reach 59 and the right button is pressed, the minutes are reset to 0.
The hours are not changed.
6. At any time in the set minutes mode, the WatchOwner can switch to the set
seconds mode by selecting the left button. To indicate this, the Watch stops blinking
the minute digits and starts blinking the second digits.
7. In the set seconds mode, the WatchOwner can advance the second digits by pressing
the right button. Each time the right button is pressed, the seconds are incremented by
one. If the seconds reach 59 and the right button is pressed, the seconds are rest to 0.
The hours and the minute digits are not changed.
Figure 4-3 SetAlarmTime use case.
Figure 4-2 SetTime use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
44 Examine the Set Ti me and Set Al ar mTi me use cases you wrote in Exercises 42 and 43 Eliminate any
redundancy by using an include relationship. Justify why an include relationship is preferable to an extend
relationship in this case.
The solution to this exercise should include three use cases: SetTime, SetAlarmTime, and a third use case (which we
called SpecifyTime) made out of the common parts of SetTime and SetAlarmTime. The essential point is to factor out
as much common functionality as possible into the third use case. The rationale is that, by eliminating redundancy,
fewer inconsistencies are introduced in the future when the use cases are modied. Below is a sample solution for this
exercise.
8. At any time in the set seconds mode, the WatchOwner can switch to the set hours
mode by pressing the left button. To indicate this, the Watch stops blinking the second
digits and starts blinking the hour digits.
9. At any time in the set hours, set minutes, and set seconds mode, the WatchOwner
can return to read time mode by pressing both Watch buttons simultaneously. The
digits stop blinking and the alarm bell stays on and stops blinking.
Exit condition 10. The Watch is in read time mode with the new alarm time set.
Special requirements None.
Use case name
SetTime
Participating actor Initiated by WatchOwner
Entry condition 1. The Watch is in read time mode.
Flow of events 2. The WatchOwner presses both Watch buttons simultaneously. The Watch enters the set
hour mode and indicates this by blinking the hour digits.
3. The SpecifyTime is included here. A the end of the SpecifyTime use case, the
WatchOwner has specied a time.
4. The WatchOwner presses both buttons simultaneously to return to the read time
mode. The new time is set.
Exit condition 5. The Watch is in read time mode with the new time set.
Special requirements None.
Figure 4-4 SetTime use case.
Use case name
SetAlarmTime
Participating actor Initiated by WatchOwner
Entry condition 1. The Watch is in read time mode.
Figure 4-5 SetAlarmTime use case.
Figure 4-3 SetAlarmTime use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
45 Assume the Fi el dOf f i cer can invoke a Help feature when lling an Emer gencyRepor t . The
Hel pRepor t Emer gency feature provides a detailed description for each eld and species which elds are
required. Modify the Repor t Emer gency use case (described in Figure 4-10) to include this help functionality.
Which relationship should you use to relate the Repor t Emer gency and Hel pRepor t Emer gency?
Flow of events 2. The WatchOwner presses both Watch buttons simultaneously. The Watch enters the set
hour mode and indicates this by blinking the hour digits. The Watch enters the set
hour mode. The WatchOwner continues to press both buttons. After one second, a
small alarm bell symbol appears on the display and blinks, indicating that the hours
of the alarm are being set (as opposed to the normal time).
3. The SpecifyTime is included here. A the end of the SpecifyTime use case, the
WatchOwner has specied a time for the alarm.
4. The WatchOwner presses both buttons simultaneously to return to the read time
mode. The new alarm time is set.
Exit condition 5. The Watch is in read time mode with the new alarm time set.
Special requirements None.
Use case name
SpecifyTime
Participating actor Initiated by WatchOwner
Entry condition 1. The Watch is in set alarm or set time mode. The hour digits are blinking.
Flow of events 2. In the set hour mode, the WatchOwner can advance the hour digits by pressing the
right button. Each time the right button is pressed, the hours are incremented by one.
If the hours reach 23 and the right button is pressed, the hours are reset to 0. The date
is not changed.
3. At any time in the set hour mode, the WatchOwner can switch to the set minutes
mode by selecting the left button. To indicate this, the Watch stops blinking the hour
digits and starts blinking the minute digits.
4. In the set minutes mode, the WatchOwner can advance the minute digits by pressing
the right button. Each time the right button is pressed, the minutes are incremented by
one. If the minutes reach 59 and the right button is pressed, the minutes are reset to 0.
The hours are not changed.
5. At any time in the set minutes mode, the WatchOwner can switch to the set
seconds mode by selecting the left button. To indicate this, the Watch stops blinking
the minute digits and starts blinking the second digits.
6. In the set seconds mode, the WatchOwner can advance the second digits by pressing
the right button. Each time the right button is pressed, the seconds are incremented by
one. If the seconds reach 59 and the right button is pressed, the seconds are rest to 0.
The hours and the minute digits are not changed.
7. At any time in the set seconds mode, the WatchOwner can switch to the set hours
mode by pressing the left button. To indicate this, the Watch stops blinking the second
digits and starts blinking the hour digits.
8. At any time in the set hours, set minutes, and set seconds mode, the WatchOwner
can return to read time mode by pressing both Watch buttons simultaneously. The
digits stop blinking and the alarm bell stays on and stops blinking.
Exit condition 9. The Watch is in read time mode.
Special requirements None.
Figure 4-6 SpecifyTime use case.
Figure 4-5 SetAlarmTime use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
The ReportEmergency use case does not need to be modied when the HelpReportEmergency is written as an
extension. The advantage of writing behavior such as help and exceptions as extensions is that they can be reused in
different use cases or in different steps of a single use case. Below is a sample solution with the <<extend>>
relationship.
46 Below are examples of nonfunctional requirements. Specify which of these requirements are veriable and
which are not.
The system must be usable. [not veriable, no precise denition of usable]
The system must provide visual feedback to the user within 1 second of issuing a command.[veriable]
The availability of the system must be above 95%. [veriable, assuming sufcient resources]
The user interface of the new system should be similar enough to the old system such that users familiar
Use case name
HelpReportEmergency
Participating actor Initiated by FieldOfficer
Entry condition 1. The Emergency Report Form is visible on the FieldOfficers screen and has not
been submitted yet. This use case extends steps 2 and 3 of ReportEmergency.
Flow of events 2. The FieldOfficer selects a eld of the Emergency Report Form (one of emergency
type, location, incident description, resource request, and hazardous
material) and presses the help key.
3. FRIEND displays in a separate window an explanatory text describing whether the eld
is required or not, what range of values is possible, and examples for typical cases.
4. The FieldOfficer may return to the form without closing the help window. For
terminals with large enough screens, the FieldOfficer can display both the form and
the help window side by side.
Exit condition 5. The use case ends when the FieldOfficer closes the help window.
Special requirements None.
Figure 4-7 HelpReportEmergency use case.
Location Use case description
Field Ofcer station 1. The FieldOfficer activates the Report Emergency function of her terminal.
2. FRIEND responds by presenting a form to the ofcer. The form includes an
emergency type menu (general emergency, re, transportation), a location, incident
description, resource request, and hazardous material elds.
3. The FieldOfficer lls the form by specifying minimally the emergency type and
description elds. The FieldOfficer may also describe possible responses to the
emergency situation and request specic resources. Once the form is completed, the
FieldOfficer submits the form by pressing the Send Report button, at which
point the Dispatcher is notied.
Dispatcher station 4. The Dispatcher is notied of a new incident report by a popup dialog. The
Dispatcher reviews the submitted information and creates an Incident in the
database by invoking the OpenIncident use case. All the information contained in
the FieldOfficers form is automatically included in the Incident. The Dispatcher
selects a response by allocating resources to the Incident (with the
AllocateResources use case) and acknowledges the emergency report by sending a
short message to the FieldOfficer.
Field Ofcer station 5. The FieldOfficer receives the acknowledgment and the selected response.
Figure 4-8 Refined description for the ReportEmergency use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
with the old system can be easily trained to use the new system. [not veriable, no precise denition of
easily trained]
47 The need for developing a complete specication may encourage an analyst to write detailed and lengthy
documents. Which competing quality of specication (see Table 4-1) may encourage an analyst to keep the
specication short?
Consistency. Long documents tend to have much redundancy and are difcult to maintain.
48 Maintaining traceability during requirements and subsequent activities is expensive, because of the addition
information that must be captured and maintained. What are the benets of traceability that outweigh this
overhead?
Ability to show that all requirements have been implemented
Ability to show that all features in the system correspond to a requirement
Ability to assess the impact of a change
Ability to track the human source of a requirement
Which of those benets are directly benecial to the analyst?
The last two benets are benecial to the analyst as it reduces the effort to change the specication without
introducing new errors.
49 Explain why multiple-choice questionnaires, as a primary means of extracting information from the user, is not
effective for eliciting requirements.
Multiple-choice questionnaires provide possible answers to the questions that are asked. This introduces two
problems: rst, the analyst must be familiar enough with the application domain to offer a good set of answers for
each question. Second, the user is constrained to select these pre-designed answers. It is then preferable that the
analyst use other elicitation methods, such as task analysis or unstructured interviews, to gather sufcient knowledge
of the application domain and to discover implicit knowledge that the user assumes everybody has. Subsequently, the
analyst can use multiple choice questionnaires to conrm a hypothesis or to prioritize certain functionality.
410 From your point of view, describe the strengths and weaknesses of users during the requirements elicitation
activity. Describe also the strengths and weaknesses of developers during the requirements elicitation activity.
This is an open-ended question. The goal is to have students realize that no one person has the complete problem/
solution under control and that it takes much work to gather all available sources of information.
Strengths of users:
they usually have detailed knowledge of the problem that needs to be solved
they have detailed knowledge of constraints imposed by the environment on the possible solutions
they have detailed knowledge of the application domain.
Weaknesses of users:
they usually have poor knowledge of the possible solutions.
they usually have poor knowledge of the formal languages for describing the problem or the solutions.
Strengths of developers:
they have detailed knowledge of different possible solutions
they have detailed knowledge of formal languages to describe the problem or the possible solutions
Weaknesses of developers:
they (initially) have poor knowledge of the problem to be solved
they can make incorrect assumptions about the problem based on their prior knowledge of different problems.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
411 Briey dene the term menu. Write your answer on a piece of paper and put it upside down on the table
together with the denitions of four other students. Compare all ve denitions and discuss any substantial
difference.
The goal of this exercise is to make students realize the ambiguity of natural language. The instructor can replace the
term menu by any other computer-oriented terms whose original meaning is still in use.
412 Write the high-level use case ManageAdver t i sement initiated by the Adver t i ser , and write detailed use cases
rening this high-level use case. Consider features that enable an Adver t i ser to upload advertisement
banners, to associate keywords with each banner, to subscribe to notices about new tournaments in specic
leagues or games, and to monitor the charges and payments made on the advertisement account. Make sure
that your use cases are also consistent with the ARENA problem statement provided in Figure 4-17.
Note: For all exercises involving writing use cases, the instructor could issue prior to the exercise writing guidelines
that the students should comply with. This will reduce the variance in the length, detail, and style of use cases,
making it easier to grade the exercise.
Figure 4-9 depicts a use case diagram illustrating a possible decomposition into a high-level use case and three
detailed use cases for ManageAdvertisement.
Figure 4-9 Detailed use cases refining the ManageAdvertisement high-level use case.
Use case name
ManageAdvertisement
Participating
actors
Initiated by Advertiser
Flow of events 1. The Advertiser may upload any number of
AdvertisementBanners, associate keywords with them, and
remove previously uploaded AdvertisementBanners. (include use
case ManageAdvertisementBanners).
2. ARENA displays the current state of the Advertisers account.
Figure 4-10 An example of a high-level the ManageAdvertisement use case.
ManageAdvertisementBanners
ManageAdvertisement
Advertiser
ManageSubscriptions
ManageAccount
initiate
include
include
include
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Figure 4-10 depicts a possible high-level use case for ManageAdvertisement. This use case should be relatively short
and simply include the detailed use cases. Details such as the attributes of the forms involved should not appear at this
level.
Figure 4-11 is an example of detailed use case for ManageAdvertisementBanners. There are many different other
possibilities given the lack of details given in the exercise. Key features should include steps to add, remove, and
modify banners, entry, and exit conditions. The use case should specify the attributes of advertisement banners that
the Advertiser can change.
413 Considering the AnnounceTour nament use case in Figure 4-24, write the event ow, entry conditions, and exit
conditions for the use case Appl yFor Tour nament , initiated by a Pl ayer interested in participating in the newly
created tournament. Consider also the ARENA problem statement provided in Figure 4-17. Write a list of
questions for the client when you encounter any alternative.
Figure 4-12 depicts a sample solution. There are, again, many different possibilities depending on the imagination of
the students. The key features of a solution include system steps that check for a number of conditions which would
make the application fail, and describe the corresponding steps. Given that this use case is relatively short, these
actions should be described in the use case as opposed to using extends relationships and many one step use cases.
The questions raised by the students should focus on when and how a Player is admitted into (and, correspondingly,
thrown out of) a league. Students could also ask about alternative ways for handling too many applications, such as
waiting lists, Players taking turns playing in different tournaments, and so on.
414 Write the event ows, entry conditions, and exit conditions for the exceptional use cases for the
AnnounceTour nament depicted in Figure 4-25. Use include relationships if necessary to remove redundancy.
This exercise should result in ve or more short use cases describing a condition under which the exception occurs
and a ow of events for handling it. Figure 4-13 is an example for one such exceptions. The student will discover that
exceptional use cases should be used for complex cases, while trivial cases (e.g., name in use, maximum number of
tournaments exceeded) could be either omitted or described within the invoking use case. This should to the
realization that writing use cases is different than writing code, as the counterpart is the user and the client, not the
computer or other developers.
3. The Advertiser may examine the details associated with his
account and any charges or payments to the account. (include use
case ManageAccount).
4. Based on the added advertisement banners and keywords, ARENA
suggests leagues and games the Advertiser may subscribe to.
5. The Advertiser may change his subscriptions to leagues and
games (include use case ManageSubscriptions).
Entry condition The Advertiser is logged into ARENA.
Exit conditions New matches can use any newly uploaded banners that match the specied keywords.
The Advertiser is notied about any new tournaments in the specied leagues or games.
Figure 4-10 An example of a high-level the ManageAdvertisement use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Use case name
ManageAdvertisementBanners
Participating
actors
Initiated by Advertiser
1. The Advertiser requests the advertisement banner area.
2. ARENA displays the list of all banners owned by this actor. The
Advertiser has the option of uploading new banners (step 3.),
editing existing banners (step 7.), or removing banners (step 11).
Flow of events 3. If the Advertiser requests the uploading of a new advertisement banner,
4. ARENA checks if the Advertiser exceeded his quote. If not, ARENA
presents the Advertiser with a form.
5. The Advertiser species an image, a set of keywords, start and end dates, and a maximum frequency.
The Advertiser can also specify that the banner can appear in any tournament.
6. ARENA updates the list of advertisement banners that can be served
for each tournament and returns to step 2.
7. If the Advertiser selects an existing banner for editing,
8. ARENA displays all the details associated with the banner
(keywords, start and end dates, maximum frequency)
9. The Advertiser species new parameters associated with the selected banners.
10. ARENA updates the list of advertisement banners that can be served
for each Tournament and returns to step 2.
11. If the Advertiser selects to view the details associated with his account,
12. ARENA displays the current total along with a list of recent charges
and payments, including at fees charged for exclusive
sponsorships and charges for each individual banners, after which
the Advertiser can select to go back to step 1.
Entry condition The Advertiser is logged into ARENA.
Exit conditions New advertisement banners can appear in new matches.
Old banners are retired immediately.
Figure 4-11 An example of a high-level the ManageAdvertisement use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Use case name
ApplyForTournament
Participating
actors
Initiated by Player
Flow of events 1. The Player requests an application for an announced tournament
2. ARENA checks if the maximum number of applications for this tournament has
been reached. If yes, the Player is notied that the tournament is full.
3. Otherwise, ARENA checks if the player is already registered with the League.
If yes, the Player is informed that his application was successful, that he will
be notied once the matches are scheduled, and the use case completes.
4. Otherwise, ARENA checks if the Player expertise rating falls within the
bounds of the tournament. If so, the Player is automatically added to the
league, and the Player is informed about the acceptance to the league and the
tournament, and that he will be notied once the matches are scheduled.
5. Otherwise, the Player is informed why he cannot be admitted into the
League.
Entry condition The Player is logged into ARENA.
Exit conditions The player is accepted in the tournament OR
The player is informed why the application failed.
Figure 4-12 ApplyForTournament use case.
Use case name
NameInUse
Participating
actors
Initiated by Player
Flow of events 1. ARENA informs the LeagueOwner that the maximum number of open
tournament for this league has been exceeded and presents a list of all the
open tournaments, sorted by date of completion
2. The AnnounceTournament use case completes.
Entry condition The maximum number of open tournaments in the current league has been exceeded and the
LeagueOwner attempts to create a new tournament.
Extends AnnounceTournament.
Exit conditions
Figure 4-13 NameInUse use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
5. Analysis: Solutions
5-1 Consider a le system with a graphical user interface, such as Macintoshs Finder, Microsofts Windows
Explorer, or Linuxs KDE. The following objects were identied from a use case describing how to copy a le
from a oppy disk to a hard disk: Fi l e, I con, Tr ashCan, Fol der , Di sk, Poi nt er . Specify which are entity
objects, which are boundary objects, and which are control objects.
Entity objects: File, Folder, Disk
Boundary objects: Icon, Pointer, TrashCan
Control objects: none in this example.
52 Assuming the same le system as before, consider a scenario consisting of selecting a le on a oppy, dragging
it to Folder and releasing the mouse. Identify and dene at least one control object associated with this
scenario.
The purpose of a control object is to encapsulate the behavior associated with a user level transaction. In this example,
we identify a CopyFile control object, which is responsible for remembering the path of the original le, the path of
the destination folder, checking if the le can be copied (access control and disk space), and to initiate the le
copying.
53 Arrange the objects listed in Exercises 5-1 and 52 horizontally on a sequence diagram, the boundary objects
to the left, then the control object you identied, and nally, the entity objects. Draw the sequence of
interactions resulting from dropping the le into a folder. For now, ignore the exceptional cases.
Figure 5-1 depicts a possible solution to this exercise. The names and parameters of the operations may vary. The
diagram, however, should at least contain the following elements:
Two boundary objects, one for the le being copied, and one of the destination folder.
At least one control object remembering the source and destination of the copy, and possibly checking for access
rights.
Two entity objects, one for the le being copied, and one of the destination folder.
In this specic solution, we did not focus on the Disk, Pointer, and TrashCan objects. The Disk object would be
added to the sequence when checking if there is available space. The TrashCan object is needed for scenarios in
which Files or Folders are deleted.
Note that the interaction among boundary objects can be complex, depending on the user interface components that
are used. This sequence diagram, however, only describes user level behavior and should not go into such details. As
a result, the sequence diagram depicts a high level view of the interactions between these object, not the actual
sequence of message sends that occurs in the delivered system.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
54 Examining the sequence diagram you produced in Exercise 54, identify the associations between these
objects.
Figure 5-2 is a possible solution for this exercises. We introduced an Item abstract class such that we can represent
common associations to Files and Folders. A consequence of this decision would be to modify the sequence
diagram of Figure 5-1 to use the Item class. The CopyFile control object has an association to the source of the Copy,
which can be either a File or a Folder. The destination of a copy is always a Folder. The Icon boundary object has
an association to the Item it represents and to the control object responsible for the copy.
55 Identify the attributes of each object that are relevant to this scenario (copying a le from a oppy disk to a hard
disk). Also consider the exception cases There is already a le with that name in the folder and There is no
more space on disk.
Figure 5-1 Sample solution for Exercise 53
Figure 5-2 Sample solution for Exercise 54
pressButton(x,y)
:CopyFile
:User
dragTo(x, y)
*dragTo(x, y)
new
releaseButton(x,y)
setSource(file)
setDestination(folder)
checkAccess()
checkAccess()
copyTo(folder)
addFile(newFile)
select()
Folder fileIcon:Icon File folderIcon:Icon
CopyFile
Folder File
Icon
I t em
source
destination
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
The scenario implies that icons have positions (since they can be moved) and that they can be selected. Each Item,
Folder or File, has a size attribute that is checked against the available space in the Disk. The exception There is
already a le with that name in the folder implies that Item names are unique within a Folder. We represent this by
adding a qualier on the relationship between Item and Folder. Finally, when copying a le, we need to copy its
contents, hence we add a contents attribute on the File object.
56 Consider the object model in Figure 5-32 in the book (adapted from [Jackson, 1995]):
Given your knowledge of the Gregorian calendar, list all the problems with this model. Modify it to correct each
of them.
The problems with Figure 5-32 are related with the multiplicity of the associations. Weeks can straddle month
boundaries. Moreover, the multiplicity on other associations can be tightened up: years are always composed of
exactly twelve months, months do not straddle year boundaries, and weeks are always composed of seven days.
Figure 5-4 depicts a possible revised model for this exercise.
57 Consider the object model of Figure 5-32 in the book. Using association multiplicity only, can you modify the
model such that a developer unfamiliar with the Gregorian calendar could deduce the number of days in each
month? Identify additional classes if necessary.
The purpose of this exercise is to show the limitation of association multiplicities. There is no complete solution to
this problem. A partial solution indicating the number of days in each month is depicted in Figure 5-5. We created
four abstract classes for each of the possible month lengths, 11 classes for each of the nonvariable months and two
classes for the month of February. The part that cannot be resolved with association multiplicities is the denition of
a leap year and that the number of days in February depends on whether the year is leap or not. In practice, this
Figure 5-3 Sample solution for Exercise 55
Figure 5-4 A naive model of the Gregorian calendar (UML class diagram).
CopyFile
Folder File
Icon I t em
position:Point
selected:Bool
size:Integer
access:ACL
name:String
contents[]:Byte
source
destination
Year
Month
Week
Day
1
12
1,2
4,5
1
7
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
problem can solved by using informal or OCL constraints, described in more detail in Chapter 9, Object Design:
Specifying Interfaces.
58 Consider a trafc light system at a four-way crossroads (e.g., two roads intersecting at right angles). Assume
the simplest algorithm for cycling through the lights (e.g., all trafc on one road is allowed to go through the
crossroad while the other trafc is stopped). Identify the states of this system and draw a statechart describing
them. Remember that each individual trafc light has three states (i.e. green, yellow, and red).
Figure 5-5 Revised class diagram indicating how many days each month includes.
Year
Month
Week
Day
1
12 1,2
4,5
1
7
FebruaryLeap MonthWith30Days MonthWith31Days FebruaryNonLeap
April
June
September
November
January
March
May
July
August
October
December
28
29
30
31
1 1 1 1
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
We model this system as two groups of lights, one for each road. Opposing lights have always the same value. Lights
from at least one group must always be red for safety reasons. We assume there are no separate lights for left and right
turns and that cycles are xed. Figure 5-6 depicts the startchart diagram for this solution.
59 From the sequence diagram Figure 2-34, draw the corresponding class diagram. Hint: Start with the
participating objects in the sequence diagram.
Figure 5-7 depicts a sample solution. The student should gure out from the sequence diagram that there should be
access paths among all three classes. One way to ensure that these paths exist is to introduce a top-level object
2Bwatch which has aggregation associations with the other three classes.
510 Consider the addition of a nonfunctional requirement stipulating that the effort needed by Adver t i ser s to
obtain exclusive sponsorships should be minimized. Change the AnnounceTour nament and the
ManageAdver t i sement use cases (solution of Exercise 4-12) so that the Adver t i ser can specify preferences in
her prole so that exclusive sponsorships can be decided automatically by the system.
Figure 5-6 Statechart diagram for simplistic traffic light system.
Figure 5-7 Sample solution for Exercise 55
after (2 seconds)
Light 1 Yellow
Light 2 Red
Light 1 Green
Light 2 Red
Light 1 Red
Light 2 Red
after (1 minute)
after (2 seconds)
Light 1 Red
Light 2 Red
Light 1 Red
Light 2 Green
Light 1 Red
Light 2 Yellow
after (1 minute)
after (2 seconds)
after (2 seconds)
2BwatchInput
pressButton1()
pressButton2()
2BwatchDisplay
digitBlinking:
blinkHours()
blinkMinutes()
blinkSeconds()
stopBlinking()
2BwatchTime
time:Date
incrementHours()
incrementMinutes()
incrementSeconds()
commitNewTime()
newTime:Date
pressButton1And2()
{none,hour,min,sec}
2Bwatch
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
A new use case, ManageSponsorships, should be written and included in the ManageAdvertisement use case and
dene the preferences that the Advertiser can specify for automatically accepting an exclusive sponsorship.
Figure 5-8 depicts a possible solution for this use case.
In the AnnounceTournament use case, steps 6 & 7 are the modied as follows:
6. The system noties the selected sponsors about the upcoming tournament and the at fee for exclusive sponsorships. For
advertisers who selected an automatic sponsorship preferences, the system automatically generates a positive answer if the fee
falls within the bounds specied by the Advertiser.
7. The system communicates their answers to the LeagueOwner.
511 Identify and write denitions for any additional entity, boundary, and control objects participating in the
AnnounceTour nament use case that were introduced by realizing the change specied in Exercise 5-10.
As specied above, this exercise should result in the identication of two new entity objects:
SponsorshipPreferences stores the time frame during which a tournament can be considered for an automatic
sponsor ship and the maximum fee the Advertiser is willing to pay without being notied. This object also has
associations to the leagues that should be considered during the automatic sponsorship decision.
ExclusiveSponsorship represents an exclusive sponsorship that was awarded to an Advertiser. It is associated
with the tournament and Advertiser objects involved in the sponsorship. It stores a boolean indicating whether
or not the sponsorship was decided automatically.
However, the instructor may choose to extend the exercise so that the students also consider the objects participating
in the ManageSponsorships use case. In this case, the student should identify a control object for the new use case
and two boundary objects for displaying the exclusive sponsorships and the preferences, respectively.
512 Update the class diagrams of Figure 5-29 and Figure 5-31 to include the new objects you identied in Exercise
5-11.
Use case name
ManageSponsorships
Participating
actors
Initiated by Advertiser
Flow of events 1. The Advertiser requests the sponsorships area..
2. ARENA displays the list of tournaments for which the Advertiser
has exclusive sponsorships and indicates for each tournament
whether the exclusive sponsorship was decided by the system or
the Advertiser.
3. The Advertiser may change the conditions under which a
tournament becomes a candidate for her exclusive sponsorship,
including specifying a set of leagues, a time frame, or a maximum
fee that she is willing to pay for an exclusive sponsorship. The
Advertiser may also select not to use the automatic sponsorship
feature and be notied about such opportunities directly.
4. ARENA records the changes of preferences for the automatic
sponsorships.
Entry condition The Advertiser is logged into ARENA.
Exit conditions ARENA checks the advertisers automatic sponsorship prole for any new tournaments created after the
completion of this use case.
Figure 5-8 ManageSponsorships use case.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Figure 5-9 depicts the objects discussed in Exercise 5-11 and their associations.
513 Draw a statechart describing the behavior of the AnnounceTour nament Cont r ol object based on the sequence
diagrams of Figures 5-26 through 5-28. Treat the sending and receiving of each notice as an event that triggers
a state change.
The AnnounceTournament is a simple linear workow. Hence, this exercise is used to test the students knowledge of
statechart diagram syntax and for choosing appropriate names for states and transitions. For modeling exercises, the
instructor is advised to select more complex workows that include decision points and concurrency, similar to the
change process described in Figure 5-22.
Figure 5-9 objects added after revising the AnnounceTournament and ManageSponsorships use case.
Figure 5-1 A UML startchart diagram for AnnounceTournamentControl.
Tournament
SponsorshipPreferences
Advertiser
1
1
*
0..1
ExclusiveSponsorship
startdate
enddate
maximum fee
League
* *
SponsorshipList
SponsorshipPreferencesForm
ManageSponsorshipControl
Create
Tournament
Wait For
Sponsor Replies
Decide on
Sponsorship
Wait for
Player Applications
notify
sponsorship deadline
expired
notify players
Decide on
Interest Groups
sponsors
notify sponsors
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
6. System Design: Decomposing the System
6-1 Decomposing a system into subsystems reduces the complexity developers have to deal with by simplifying the
parts and increasing their coherence. Decomposing a system into simpler parts usually results into increasing a
different kind of complexity: Simpler parts also means a larger number of parts and interfaces. If coherence is
the guiding principle driving developers to decompose a system into small parts, which competing principle
drives them to keep the total number of parts small?
Decreasing coupling is the principle that competes with increasing coherence. A large number of parts will result in a
large number of interfaces and many dependencies among the parts.
62 In Section 6.4.2 in the book, we classied design goals into ve categories: performance, dependability, cost,
maintenance, and end user. Assign one or more categories to each of the following goals:
Users must be given a feedback within 1 second after they issue any command. [Performance]
The Ti cket Di st r i but or must be able to issue train tickets, even in the event of a network failure.
[Dependability]
The housing of the Ti cket Di st r i but or must allow for new buttons to be installed in the event the
number of different fares increases. [Maintenance]
The Aut omat edTel l er Machi ne must withstand dictionary attacks (i.e., users attempting to discover a
identication number by systematic trial). [Dependability]
The user interface of the system should prevent users from issuing commands in the wrong order. [End
user]
63 You are developing a system that stores its data on a Unix le system. You anticipate that you will port future
versions of the system to other operating systems that provide different le systems. Propose a subsystem
decomposition that anticipates this change.
A Bridge pattern should be used to dene a generic interface to the storage subsystem. Consequently, different
implementations of the storage subsystem can be provided to dealing with specic le systems. Note that the Bridge
pattern does not address the problem of moving les from one le system to another. To address this problem, a
separate conversion utility needs to be developed.
64 Older compilers were designed according to a pipe and lter architecture, in which each stage would transform
its input into an intermediate representation passed to the next stage. Modern development environments,
including compilers integrated into interactive development environments with syntactical text editors and
source-level debuggers, use a repository architecture. Identify the design goals that may have triggered the shift
from pipe and lter to repository architecture.
Older compilers trace their lineage to batch systems, in which a stack of cards were fed to the computer by system
administrators and developers would pick up the resulting printout later in the day. The design goal of such system
was high throughput, so that as many jobs as possible could be executed. A pipe and lter architecture allows
parallelism between subsequent stages in the pipeline and achieves this design goal.
Modern interactive development environments are interactive, compilation occurs incrementally as the user types.
The design goal of such environments is a short response time.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
65 Consider the model/view/control example depicted in Figures 6-16 and 6-15.
a. Redraw the collaboration diagram of Figure 6-16 as a sequence diagram.
b. Discuss how the MVC architecture helps or hurts the following design goals.
Extensibility (e.g., the addition of new types of views).
The MVC helps extensibility in terms of new types of views, since neither model objects nor existing view objects
need to be modied to accommodate the new view.
Response time (e.g., the time between a user input and the time all views have been updated)
A pure MVC hurts response time, because the view the user sees is updated only after the model has been updated.
This can be an issue for commands involving a rapid series of interactions (e.g., dragging or resizing a geometrical
shape).
Modiability (e.g., the addition of new attributes in the model)
The MVC helps this goal, since only the views that need to be aware of the new attribute need to be modied. All
other views can be left unchanged.
Access control (i.e., the ability to ensure that only legitimate users can access specic parts of the model).
The MVC helps this goal, as the model is accessed using a clear interface (the public methods of the model objects)
which can be controlled, for example, using a proxy pattern for each model class.
66 List design goals that would be difcult to meet when using a closed architecture with many layers, such as the
OSI example depicted in Figure 6-10.
Figure 6-1 Sample solution for Exercise 6-5.
openFolder()
:InfoView
:User
viewFolderInfo()
enterNewFileName(file,name)
new
notifySubscribedView()
subscribeToFileEvents()
:Controller file:Model
:FolderView
new
subscribeToFileEvents()
setName(name)
notifySubscribedView()
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
A closed architecture requires method invocations across each layer boundary and often also copying data between
each layer. This hurts both response time and memory foot print.
67 In many architectures, such as the three- and four-tier architectures (Figures 6-21 and 6-22), the storage of
persistent objects is handled by a dedicated layer. In your opinion, which design goals have lead to this
decision?
Portability: Separating the application from the details of storage allows the storage layer to be replaced by a different
one (e.g., switching database management systems).
Robustness: It is easier to guarantee the integrity of the stored data if it is handled by a single layer.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
7. System Design: Addressing Design Goals: Solutions
7-1 Consider a system that includes a Web server and two database servers. Both database servers are identical:
The rst acts as a main server, while the second acts as a redundant backup in case the rst one fails. Users use
Web browsers to access data through the Web server. They also have the option of using a proprietary client that
accesses the databases directly. Draw a UML deployment diagram representing the hardware/software
mapping of this system.
This exercise tests the students knowledge of deployment diagrams. Several different solutions exist depending on
the interpretation of the word server. Given that the system uses a hot backup redundancy to increase availability,
the student can infer that high availability is a design goal for the system. In this case, the system designers would
also assign each server to a dedicated host, rst, to increase the independence of failures, second, to limit damage in
case of failure. Figure 7-1 depicts such a solution with three servers and two clients. An instructor may want to accept
a solution where both database servers and the Web server run on the same host, given that this exercise tests
knowledge about deployment diagrams, not about dependable systems.
72 Consider a legacy, fax-based, problem-reporting system for an aircraft manufacturer. You are part of a
reengineering project replacing the core of the system by a computer-based system, which includes a database
and a notication system. The client requires the fax to remain an entry point for problem reports. You propose
an E-mail entry point. Describe a subsystem decomposition, and possibly a design pattern, which would allow
both interfaces.
There are two main approaches to this problem. The rst approach is to design the system with an E-mail frontend
and add a subsystem which converts fax problem reports to E-mail (possibly with the assistance of an actor). The E-
mail front end then becomes a single point of entry into the problem reporting system. The second approach (depicted
in Figure 7-2) is to provide a network level interface to the problem reporting system (e.g., using RPC, RMI,
CORBA, or plain sockets) and to build two different clients to the system, one processing E-mail, the other
processing fax reports. The rst solution is cheaper to realize and would be appropriate if the fax front end is short
lived or no other frontends are planned. The second solution is more general and allows the addition of other
frontends in the future. Design goals dened during system design would be used to decide between these two
Figure 7-1 Sample solution for Exercise 7-1.
:WebBrowser
:WebServer
:ClientHost :WebServerHost
:DBClient
:ClientHost
backup:Database
:DBServerHost
main:Database
:DBServerHost
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
alternatives. The main point the student should realize is that there are alternative solutions to subsystem
decomposition issues and that design goals can be used to sort them out.
73 You are designing the access control policies for a Web-based retail store. Customers access the store via the
Web, browse product information, input their address and payment information, and purchase products.
Suppliers can add new products, update product information, and receive orders. The store owner sets the retail
prices, makes tailored offers to customers based on their purchasing proles, and provides marketing services.
You have to deal with three actors: St or eAdmi ni st r at or , Suppl i er , and Cust omer . Design an access control
policy for all three actors. Cust omer s can be created via the Web, whereas Suppl i er s are created by the
St or eAdmi ni st r at or .
An access control policy is represented with a matrix. The columns represent objects whose access is controlled, the
rows represent the actors accessing the objects, the cells contain the operations that an actor is allowed to invoke for a
specic object. In this exercise, there are four objects under access control: the product (including product
information and price), the customer information, the supplier information, and the order. There are four actors which
should be taken into account: the three actors mentioned in the exercise and the unregistered web user which can
browse the product catalog and the create a new customer (as indicated in the last sentence of the exercise. Table 7-1
depicts a possible access matrix for the text above. The name of the operations may be different from one solution to
another. The instructor may consider correct solutions which merge the unregistered user row with the customer row.
Figure 7-2 Sample solution for Exercise 7-2.
Table 7-1 Access control matrix for the web store of exercise 6.5.
Product CustomerInfo SupplierInfo Order
Unregistered web
user
getInfo() create()
Customer
getInfo()
getPrice()
updateInfo() create()
Supplier
create()
getInfo()
updateInfo()
updateInfo() process()
StoreAdministrator
updatePrice() verifyInfo()
examineProfile()
create() examine()
EmailFrontEnd
FaxFrontEnd
ProblemReporting
StorageSubsystem
NotificationSubsystem
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
74 Select a control ow mechanism you nd most appropriate for each of the following systems. Because multiple
choices are possible in most cases, justify your choices.
a Web server designed to sustain high loads [Threaded control ow]
a graphical user interface for a word processor [Event-based control ow]
a real-time embedded system (e.g., a guidance system on a satellite launcher) [Event-based control ow
(e.g., preemptive, interrupt driven, scheduler)]
75 Why are use cases that describe boundary conditions described during system design (as opposed to during
requirements elicitation or analysis)?
Use cases that describe boundary conditions depend on system design decisions. For example, software architecture
decisions need to be made before developers can describe how the system is started or shutdown.
76 You are designing a caching subsystem that temporarily stores data retrieved over the network (e.g., web pages)
into a faster access storage (e.g., the hard disk). Due to a change in requirements, you dene an additional
service in your subsystem for conguring cache parameters (e.g., the maximum amount of hard disk the cache
can use). Which project participants do you notify?
You should notify at least:
all teams using your subsystem
the team responsible for testing your subsystem
the team responsible for documenting the interface of your subsystem
the architecture team
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
8. Object Design: Reusing Pattern Solutions
8-1 Consider the ARENA object design model. For each of the following objects, indicate if it is an application object
or a solution object:
League [application object]
LeagueStore [solution object]
LeagueXMLStoreImplementor [solution object]
Match [application object]
MatchView [solution object]
Move [application object]
ChessMove. [application object]
82 Indicate which occurrences of the following inheritance relationships are specication inheritance and which
are implementation inheritance:
A Rect angl e class inherits from a Pol ygon class. [specication inheritance]
A Set class inherits from a Bi nar yTr ee class. [implementation inheritance]
A Set class inherits from a Bag class (a Bag is dened as an unordered collection). [specication
inheritance]
A Pl ayer class inherits from a User class. [specication inheritance]
A Wi ndow class inherits from a Pol ygon class. [implementation inheritance]
83 Consider an existing game of bridge written in Java. We are interested in integrating this bridge game into
ARENA. Which design pattern would you use? Draw a UML class diagram relating the ARENA objects with some
of the classes you would expect to nd in the bridge game.
Integrating an existing game requires writing one Adapter per ARENA class in the Game interface. Figure 8-1 depicts an
example for the GameMove class.
84 Consider a workow system supporting software developers. The system enables managers to model the
process the developers should follow in terms of processes and work products. The manager can assign specic
processes to each developer and set deadlines for the delivery of each work product. The system supports
several types of work products, including formatted text, picture, and URLs. The manager, while editing the
workow, can dynamically set the type of each work product at run time. Assuming one of your design goals is
to design the system so that more work product types can be added in the future, which design pattern would
you use to represent work products?
We use a bridge pattern. The interface class stores the deadline and type of work product and provides the interface to
the rest of the application. The implementor classes store the content of the work product and can be substituted at run
time. All implementor classes comply to the same abstract Wor kPr oduct Cont ent interface.
Figure 8-1 Sample solution for 83.
GameMove
execut e( )
adaptee
BridgeMove
BridgeMoveAdapter
execute()
Array
play()
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
85 Consider a system that includes a database client and two redundant database servers. Both database servers
are identical: the rst acts as a main server, the second acts as a hot back-up in case the main server fails. The
database client accesses the severs through a single component called a gateway, hence hiding from the
client which server is currently being used. A separate policy object called a watchdog monitors the requests
and responses of the main server and, depending on the responses, tells the gateway whether to switch over to
the back-up server. What do you call this design pattern? Draw a UML class diagram to justify your choice.
This can be interpreted as a Strategy in which two concrete strategies are identical, that is, they are instances of the
same class (Figure 8-3). It can also be argued that this is a Proxy pattern which interacts with an external policy
object.
86 In Section 8.4.1, we used a Bridge pattern to decouple the implementation of the ARENA LeagueStore subsystem
from its interface, enabling us to provide different implementations for the purpose of testing. Ideally, we would
apply the Bridge pattern to each subsystem in our system design to facilitate testing. Unfortunately, this is not
always possible. Give an example of a subsystem where the Bridge pattern cannot be used.
A Bridge pattern can be used in the case of a subsystem with a single Facade. Subsystems that offer a more complex
interface, for example a whitebox framework used for realizing a user interface, are difcult to encapsulate.
87 Consider the following design goals. For each of them, indicate the candidate pattern(s) you would consider to
satisfy each goal:
Given a legacy banking application, encapsulate the existing business logic component. [Adapter]
Given a chess program, enable future developers to substitute the planning algorithm that decides on the
next move with a better one. [Bridge]
Given a chess program, enable a monitoring component to switch planning algorithms at runtime, based
on the opposing players style and response time. [Strategy]
Given a simulation of a mouse solving a maze, enable the path evaluation component to evaluate different
paths independently of the types of moves considered by the mouse. [Command]
Figure 8-2 Sample solution for Exercise 84.
Figure 8-3 Sample solution for Exercise 85.
PictureContent
Wor kPr oduct Cont ent WorkProduct
TextContent URLContent
WorkflowSystem
deadline:Date
type:String
Gateway Dat abaseSer ver
query() query()
Client
Watchdog
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
88 Consider an application that must select dynamically an encryption algorithm based on security requirements
and computing time constraints. Which design pattern would you select? Draw a UML class diagram depicting
the classes in the pattern and justify your choice.
A strategy pattern can be used to address this problem. The EncryptionAlgorithm interface denes the common
interface that all encryption algorithms must comply with. A Policy class selects a concrete algorithm based on the
computing time constraints and the security requirements of the application. The Client interacts only with the
Context interface which stores the reference to the concrete algorithm.
Figure 8-4 Sample solution for Exercise 88.
Context Encr ypt i onAl gor i t hm
encrypt(txt)
ConcreteAlgorithm1 ConcreteAlgorithm2
encrypt(txt)
Client
Policy
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
9. Object Design: Specifying Interfaces
9-1 Consider the List interface in the java.util package for ordered collections of objects. Write preconditions
and post conditions in OCL for the following operations:
The constraints below assume that an ordered association between the list and its contained elements is called
elements.
int size() returns the number of elements in the list.
context List::size() post: result = elements->size
void add(Object e) adds an object at the end of the list.
context List::add(e) post: contains(e) and size() = @pre.size() + 1
void remove(Object e) removes an object from the end of the list.
context List::remove(e) post:
(@pre.contains(e) implies size() = @pre.size() - 1) and
(not @pre.contains(e) implies size() = @pre.size()) and
result = @pre.contains(e)
boolean contains(Object e) returns true if the object is contained in the list.
context List::contains(e) post: result = elements->includes(e)
Object get(int idx) returns the object located at index idx, 0 being the index of the rst object in the
list.
context List::get(idx) pre: idx >= 0 and idx < size()
context List::get(idx) post: result = elements->at(idx+1)
92 Consider the Set interface in the java.util package. Write preconditions and post conditions in OCL for the
following operations:
The constraints below assume that the association between the list and its contained elements is called elements.
int size() returns the number of elements in the set.
context Set::size() post: result = elements->size()
void add(Object e) adds an object to the set. If the object is already in the set, does nothing.
context Set::add(e) post:
contains(e) and
(@pre.contains(e) implies size() = @pre.size()) and
(not @pre.contains(e) implies size() = @pre.size()+1)
void remove(Object e) removes an object from the set
context Set::remove(e) post:
not contains(e) and
(@pre.contains(e) implies size() = @pre.size() - 1) and
(not @pre.contains(e) implies size() = @pre.size())
boolean contains(Object e) returns true if the object is contained in the set.
context Set::contains(e) post: elements->includes(e)
93 Consider the Collection interface in the java.util package, which is the ancestor of both List and Set.
Write preconditions and postconditions for the operations below and modify the constraints you wrote in
exercises 9-1 and 92, knowing that contracts are inherited. Make sure you comply with the Liskov Substitution
Principle.
int size() returns the number of elements in the collection.
context Collection::size() post: result = elements->size()
void add(Object e) adds an object to the collection.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
context Collection::add(e) post: contains(e)
void remove(Object e) removes an object from the collection.
context Collection::remove(e) post:
(@pre.contains(e) implies size() = @pre.size() - 1) and
(not @pre.contains(e) implies size() = @pre.size())
boolean contains(Object e) returns true if the object is in the collection.
context Collection::contains(e) post: elements->includes(e)
With contract inheritance, no constraints on remove, contains, and size of List and Set need to be added. The post
condition on add() on Set and List become:
context List::add(e) post: size() = @pre.size() + 1
context Set::add(e) post:
(@pre.contains(e) implies size() = @pre.size()) and
(not @pre.contains(e) implies size() = @pre.size() + 1)
94 Consider a Rectangle class and a Square class that inherits from the Rectangle class:
Write post conditions for the Rectangle.setWidth(w:int) and the Rectangle.setHeight(h:int)
operations in terms of the Rectangle.getWidth():int and the Rectangle.getHeight():int
operations.
context Rectangle::setWidth(w) post:
w = getWidth() and getHeight() = @pre.getHeight()
context Rectangle::setHeight(h) post:
h = getHeight() and getWidth() = @pre.getWidth()
Write an invariant for the Square class stating that the width and height of a Square should always be the
same.
context Square inv: getWidth() = getHeight()
Consider the rules for inheriting contracts described in Section 9.4.5 in the context of the
Square.setWidth() and Square.setHeight() operations of the Square class. Are all rules met? Why
not? What should change in the model?
As written above, the Square.setWidth() and Square.setHeight() would violate the postconditions of the
Rectangle.setWidth() and the Rectangle.setHeight(). Square should not inherit from Rectangle as it does not
provide the same external behavior with respect to setWidth() and setHeight() (as dened by the LSP).
To make this exercise clearer, the instructor could write the postcondition for Rectangle.getWidth(), as the second
part of the postcondition is often missed.
95 Consider a sorted list. Write an invariant in OCL denoting that the elements of the list are sorted.
The key to this exercise is to realize that the List class in Java provides an indexOf(element) method.
context SortedList inv:
elements->forAll(e1, e2:Object|
e1.lessThan(e2) implies indexOf(e1) < indexOf(e2))
96 Consider a sorted binary tree data structure for storing integers. Write invariants in OCL denoting that
All nodes in the left subtree of any node contain integers that are less than or equal to the current node, or
the subtree is empty.
context BinaryTree inv:
left.getSubElements()->size = 0 or
left.getSubElements()->forAll(e:Integer| e <= getElement())
All nodes in the right subtree of any node contain integers that are greater than the current tree, or the
subtree is empty.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
context BinaryTree inv:
right.getSubElements()->size = 0 or
right.getSubElements()->forAll(e:Integer|e > getElement())
The tree is balanced.
context BinaryTree inv:
left.getSubElements()->size <= right.getSubElements()->size + 1 and
right.getSubElements()->size <= left.getSubElements()->size + 1
97 Consider a simple intersection with two crossing roads and four trafc lights. Assume a simple algorithm for
switching lights, so that the trafc on one road can proceed while the trafc on the other road is stopped. Model
each trafc light as an instance of a Tr af f i cLi ght class with a st at e attribute that can be either r ed, yel l ow,
or gr een. Write invariants in OCL on the st at e attribute of the Tr af f i cLi ght class that guarantee that the
trafc cannot proceed on both roads simultaneously. Add associations to the model to navigate the system, if
necessary. Note that OCL constraints are written on classes (as opposed to instances).
We add an association called right to denote the trafc light on the road to the right. Hence, the expression
right.right denotes the opposing light.
/ * Opposi ng l i ght s have al ways t he same st at e * /
context TrafficLight inv:
state = right.right.state
/ * I f t he cur r ent l i ght i s gr een or yel l ow, t he l i ght on t he r i ght must be r ed * /
context TrafficLight inv:
(state = #green or state = #yellow) implies right.state = #red
98 After reading Sections 9.6.2 and 9.6.3, write constraints for a RoundRobi nSt yl e class and a RoundRobi nRound
class, implementing the Tour nament St yl e and the Round interfaces, respectively. Assume that
RoundRobi nSt yl e plans a series of Rounds so that each Pl ayer is paired with the other Pl ayer s exactly once in
the Tour nament . Note that the number of Rounds depends on whether the number of Pl ayer s in the Tour nament
is odd or even, and that a given Pl ayer cannot play more than once in a given Round.
/ * I f t he number of pl ayer s i s even, t he number of r ounds i s one l ess t han t he
* number of pl ayer s. Ot her wi se, i t i s equal t o t he number of pl ayer s * /
context RoundRobinStyle::planRounds(t:Tournament) post:
if (t.players.size.mod(2) = 0) then
result->size = t.players.size - 1
else
result->size = t.players.size
endif
/ * Each pl ayer pl ays each ot her i n a RoundRobi nSt yl e t our nament .
context RoundRobinStyle::planRounds(t:Tournament) post:
t.players->forAll(p:Player|
p.getMatches(t).players->includesAll(t.players))
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
10. Mapping Models to Code: Solutions
10-1 In Web pages, tables consist of rows, which in turn consist of cells. The actual width and height of each cell is
computed based in part on its content (e.g., the amount of text in the cell, the size of an image in the cell), and
the height of a row is the maximum of the heights of all cells in the row. Consequently, the nal layout of a table
in a Web page can only be computed once the content of each cell has been retrieved from the Internet. Using
the proxy pattern described in Figure 10-7, describe an object model and an algorithm that would enable a Web
browser to start displaying a table before the size of all cells is known, possibly redrawing the table as the
content of each cell is downloaded.
The proxy design pattern is applied to the table cell. When created, TableCellProxy estimates its dimensions as best
as it can and updates them as the content is retrieved. Every time TableCellProxy provides a new estimate of its
dimensions, it signals Table (e.g., via a semaphore or a listener). The Table then recomputes the dimensions and
positions of all of its rows and redraws itself.
In general, the challenge of this type of approach is not the algorithm but to provide a visual behavior that does not
confuse the user, especially when only part of the table is visible.
102 Apply the appropriate transformations described in Section 10.4.2 to the associations below. Assume that all
associations are bidirectional and that they can change during the lifetime of each object. Write the source code
needed to manage the associations, including class, eld, and method declarations, method bodies, and
visibility.
Figures 10-2 through 10-5 depict sample classes for this exercise. Import statements and constructors were omitted.
Since no ordering was specied on the associations, a Set was used to represent the many side. If the student selects a
List, additional checking code must be added to ensure that multiple associations to the same object are not created.
Figure 10-1 Sample class diagram for Exercise 10-1.
TableCell
x,y,w,h:int
RealTableCell TableCellProxy
1 0..1
paint()
paint() paint()
TableRow
Table
1
*
1
*
x,y,w,h:int
paint()
x,y,w,h:int
paint()
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
public class Mailbox {
private Set folders = new HashSet();
public Folder [] getFolders() {
return (Folder[])folders.toArray();
}
public void addFolder(Folder f) {
folders.add(f);
f.setMailbox(this);
}
public void removeFolder(Folder f) {
folders.remove(f);
f.setMailbox(null);
}
}
Figure 10-2 Class Mailbox.
public class Folder {
private Mailbox mailbox = null;
private Set messages = new HashSet();
public Mailbox getMailbox() {
return mailbox;
}
public void setMailbox(Mailbox newMailbox) {
if (newMailbox != mailbox) {
Mailbox oldMailbox = mailbox;
mailbox = newMailbox;
if (newMailbox != null) {
newMailbox.addFolder(this);
}
if (oldMailbox != null) {
oldMailbox.removeFolder(this);
}
}
}
public Message [] getMessages() {
return (Message[])messages.toArray();
}
public void addMessage(Message m) {
messages.add(m);
m.setFolder(this);
}
public void removeMessage(Message m) {
messages.remove(m);
m.setFolder(null);
}
}
Figure 10-3 Class Folder
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
public class Message {
private Folder folder = null;
private Set views = new HashSet();
public Folder getFolder() {
return folder;
}
public void setFolder(Folder newFolder) {
if (newFolder != folder) {
Folder oldFolder = folder;
folder = newFolder;
if (newFolder != null) {
newFolder.addMessage(this);
}
if (oldFolder != null) {
oldFolder.removeMessage(this);
}
}
}
public View [] getViews() {
return (View[])views.toArray();
}
public void addView(View v) {
if (!views.contains(v)) {
views.add(v);
v.addMessage(this);
}
}
public void removeView(View v) {
if (views.contains(v)) {
views.remove(v);
v.removeMessage(this);
}
}
}
Figure 10-4 Class Message.
public class View {
private Set messages = new HashSet();
public Message [] getMessages() {
return messages.toArray();
}
public void addMessage(Message m) {
if (!messages.contains(m)) {
messages.add(m);
m.addView(this);
}
}
public void removeMessage(Message m) {
if (messages.contains(m)) {
messages.remove(m);
m.removeView(this);
}
}
}
Figure 10-5 Class View.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
103 Apply the appropriate transformations described in Section 10.4.2 to the associations below. Assume that all
associations are bidirectional, but that the aggregation associations do not change after each object has been
created. In other words, the creator of each class must be modied so that aggregations are initialized during
the creation of each object. Write the source code needed to manage the associations, including class, eld, and
method declarations, method bodies, and visibility.
Figures 10-6 through 10-9 depict a sample solution. The constructor of Tournament and Round set the one side of the
aggregation association, removing the need for a setLeague and setTournament method.
public class League {
private Set tournaments;
public League() {
tournaments = new HashSet();
}
public Tournament [] getTournaments() {
return (Tournament[])tournaments.toArray();
}
public void addTournament(Tournament t) {
tournaments.add(t);
}
public void removeTournament(Tournament t) {
tournaments.remove(t);
}
}
Figure 10-6 Class League.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
public class Tournament {
private League league;
private Set rounds;
private Set players;
public Tournament(League l) {
rounds = new HashSet();
players = new HashSet();
league = l;
l.addTournament(this);
}
public League getLeague() {
return league;
}
public Round [] getRounds() {
return (Round[])rounds.toArray();
}
public void addRound(Round r) {
rounds.add(r);
}
public void removeRound(Round r) {
rounds.remove(r);
}
public Player [] getPlayers() {
return (Player[])players.toArray();
}
public void addPlayer(Player p) {
if (!players.contains(p)) {
players.add(p);
p.addTournament(this);
}
}
public void removePlayer(Player p) {
if (players.contains(p)) {
players.remove(p);
p.removeTournament(this);
}
}
}
Figure 10-7 Sample class Tournament.
public class Round {
private Tournament tournament;
public Round(Tournament t) {
tournament = t;
t.addRound(this);
}
public Tournament getTournament() {
return tournament;
}
}
Figure 10-8 Class Round.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
104 Figure 10-15 depicts the checking code for the addPlayer() method of Tournament. Write the checking code
for the other constraints associated with Tournament depicted in Figure 9-16.
Figure 10-10 depicts the checking code for Tournament.removePlayer(). The key points are that
preconditions are checked rst,
some values may have to be saved for checking postconditions before the real work is done,
postconditions and invariants are checked after the real work is done, and
the conditions checked (i.e., the conditions under which an exception should be raised) are the opposite of the
conditions stated in the contracts (i.e., the conditions that should always be true).
public class Player {
private Set tournaments;
public Player() {
tournaments = new HashSet();
}
public Tournament [] getTournaments() {
return (Tournament[])tournaments.toArray();
}
public void addTournament(Tournament t) {
if (!tournaments.contains(t)) {
tournaments.add(t);
t.addPlayer(this);
}
}
public void removeTournament(Tournament t) {
if (tournaments.contains(t)) {
tournaments.remove(t);
t.removePlayer(this);
}
}
}
Figure 10-9 Class Player.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
105 Write checking code for the contracts of Tour nament St yl e and Round described in Section 9.6.2. Write
checking code for preconditions, postconditions, and invariants
Figures 10-11 and 10-12 depict the checking code associated with the constraints in Section 9.6.2.
public class Tournament {
//...
private List players;
public void removePlayer(Player p) throws UnknownPlayer, KnownPlayer,
IllegalNumPlayers, IllegalLeague, IllegalName
{
// check precondition isPlayerAccepted(p)
if (!isPlayerAccepted(p)) {
throw new UnknownPlayer(p);
}
// check precondition getNumPlayers() < maxNumPlayers
if (getNumPlayers() == getMaxNumPlayers()) {
throw new TooManyPlayers(getNumPlayers());
}
// save values for postconditions
int pre_getNumPlayers = getNumPlayers();
// accomplish the real work
players.add(p);
p.addTournament(this);
// check post condition !isPlayerAccepted(p)
if (isPlayerAccepted(p)) {
throw new KnownPlayer(p);
}
// check post condition getNumPlayers() = @pre.getNumPlayers() - 1
if (getNumPlayers() != pre_getNumPlayers - 1) {
throw new IllegalNumPlayers(getNumPlayers());
}
// check invariant maxNumPlayers > 0
if (getMaxNumPlayers() <= 0) {
throw new IllegalMaxNumPlayers(getMaxNumPlayers());
}
// check invariant getNumPlayers() <= getMaxNumPlayers()
if (getNumPlayers() > getMaxNumPlayers()) {
throw new IllegalNumPlayers(getNumPlayers());
}
// check invariant getLeague() != null
if (getLeague() == null) {
throw new IllegalLeague();
}
// check invariant getName() != null
if (getName() == null) {
throw new IllegalName();
}
}
//...
}
Figure 10-10 Checking code for removePlayer()
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
.
public class TournamentStyle {
// ...
public List planRounds(Tournament t) throws IllegalTournament,
TournamentAlreadyPlanned,PlayerNotAssigned, OverlappingRounds
{
// check precondition t <> nil
if (t == null) {
throw new IllegalTournament();
}
// check precondition t.rounds = nil
if (t.getRounds() != null) {
throw new TournamentAlreadyPlanned();
}
// check precondition legalNumPlayers(t.players->size))
if (!legalNumPlayers(t.getNumPlayers())) {
throw new IllegalNumPlayers();
}
// ... do the real work ...
List rounds = new ArrayList();
// ....
// check postcondition
// t.getPlayers()->forAll(p|p.getMatches(t)->notEmpty)
for (Iterator i = t.getPlayers().iterator(); i.hasNext();) {
Player p = (Player)i.next();
if (p.getMatches() == null || p.getMatches().size() == 0) {
throw new PlayerNotAssigned(p);
}
}
// check postcondition result->forAll(r1, r2| r1<>r2 implies
// r1.getEndDate().before(r2.getStartDate()) or
// r1.getStartDate().after(r2.getEndDate())
for (Iterator i = rounds.iterator(); i.hasNext();) {
Round r1 = (Round)i.next();
for (Iterator j = rounds.iterator(); j.hasNext();) {
Round r2 = (Round)j.next();
if (r1 != r2) {
if (!r1.getEndDate().before(r2.getStartDate()) and
!r1.getStartDate().after(r2.getEndDate())) {
throw new OverlappingRounds(r1, r2);
}
}
}
}
return rounds;
}
}
Figure 10-11 Checking code for TournamentStyle.planRounds().
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
public class Round {
List matches = new ArrayList();
boolean planned = false;
boolean completed = false;
public void plan() throws RoundNotPlanned, PlayerInMoreThanOneMatchInRound {
previousRoundCompleted = getPreviousRound().isCompleted();
// ... plan the round;
// check postcondition
if (previousRoundCompleted) {
if (!isPlanned()) {
throw RoundNotPlanned();
}
}
// check invariant
for (Iterator i = matches.iterator(); i.hasNext();) {
Match m1 = (Match)i.next();
for (Iterator j = m1.getPlayers().iterator(); j.hasNext();) {
Player p = (Player)j.next();
for (Iterator k = p.getMatches(); k.hasNext();) {
Match m2 = (Match)k.next();
if (m1 != m2) {
if (m1.getRound() == m2.getRound()) {
throw PlayerInMoreThanOneMatchInRound(m1, m2);
}
}
}
}
}
}
public boolean isPlanned() throws IllegalNumPlayersInMatch {
// check postcondition
if (planned) {
for (Iterator i = matches.iterator(); i.hasNext();) {
Match m = (Match)i.next();
if (m.getPlayers()->size() !=
getTournament().getLeague.getGame().getNumPlayersPerMatch) {
throw IllegalNumPlayersInMatch(m);
}
}
}
return planned;
}
public boolean isCompleted() throws NoWinnerInCompletedMatch {
// check postcondition
if (completed) {
for (Iterator i = matches.iterator(); i.hasNext();) {
Match m = (Match)i.next();
if (m.getWinner() == null) {
throw NoWinnerInCompletedMatch();
}
}
}
return completed;
}
}
Figure 10-12 Checking code for Round.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
106 Design a relational database schema for the object model of Figure 10-30. Assume Leagues, Tournaments,
Players, and Rounds have a name attribute and a unique identier. Additionally, Tournaments and Rounds have
start and end date attributes. When different transformations are available, explain the trade-off involved.
Figure 10-13 depicts a sample solution for this exercise. This exercise checks only a basic understanding of the class
model to relational schema since most of the tables in this schema already appear in various gures in Chapter 10.
The next exercise focuses more on modeling.
107 Draw a class diagram representing the application domain facts below, and map it to a relational schema.
A project involves a number of participants.
Participants can take part in a project either as project manager, team leader, or developer.
Within a project, each developer and team leader is part of at least one team.
A participant can take part in many projects, possibly in different roles. For example, a participant can be
a developer in project A, a team leader in project B, and a project manager in project C. However, the
role of a participant within a project does not change.
Figure 10-14 depicts a class diagram for the above domain facts and Figure 10-15 depicts the corresponding schema.
Worth of note:
The key to the exercise is to create a Role class between the participant and the project.
The association between Manager and Project is not explicit in the above facts but should be known to the
student from the denition of project manager in the book. As a one-to-one association, it can be represented as a
buried association in either the Manager or the Project table.
Also, implicit in the above description is that each participant takes part in a project in exactly one role, which
would have to be represented with an OCL constraint and checking code. This sample solution does not take into
account this constraint in the sense that it would allow a single person to take part in the same project in multiple
roles.
Figure 10-13 Sample solution for Exercise 106.
player:long
TournamentPlayerAssociation table
tournament:long
id:long
League table
name:text[25]
id:long
Player table
name:text[25]
id:long
Tournament table
name:text[25] start:date end:date
id:long
Round table
name:text[25] start:date end:date
league:long
tournament:long
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
108 There are two general approaches for mapping an association to a set of collections. In Section 10.6.2, we map
the N-ary association Statistics to two classes, a simple Statistics class to store the attributes of the
association, and a StatisticsVault class to store the state of the links among the association links. In Section
10.4.2, we described an alternative approach where the association links are stored in one or both classes at
the ends of the association. In the event associations were stored in both classes, we added mutually recursive
methods to ensure that both data structures remained consistent. Use this second approach to map the N-ary
Statistics association to Collections. Discuss the trade-offs you encounter and the relative advantages of
each approach.
Mapping the N-ary Statistics association to a set of binary associations which are in turn mapped to a set of
collections results in a loss of information and a lot of redundancy. In Figure 10-25, there are only three methods for
retrieving statistics objects based on three possible pairs of objects. When mapping binary associations to collections,
there will be at least two different methods for each case provided by two different classes. This distribution of code
makes it more difcult to maintain it and to extend it when new classes are added to the N-ary association.
The only advantage of the binary association approach is that it is mechanical and could be supported by a code
generation tool. While this would take care of ensure that redundant code is realized correctly, it does not address the
problem of extensibility.
Figure 10-14 Class diagram for Exercise 107.
Figure 10-15 Relational schema for Exercise 107.
Developer TeamLeader Manager
Rol e Project Person
Team
*
*
*
*
1
*
1
1
1
1
1
team:long
DeveloperTeamAssociation table
developer:long
id:long
Role table
project:long id:long
Developer table
person:long
id:long
Team table
project:long leader:long
id:long
TeamLeader table
id:long
Manager table
manager:long
Project table
id:long id:long
Person table
*
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
11. Testing: Solutions
11-1 Correct the faults in the i sLeapYear ( ) and get NumDaysI nMont h( ) methods of Figure 11-12 and generate test
cases using the path testing methods. Are the test cases you found different than those of Table 11-14 and Figure
11-13? Why? Would the test cases you found uncover the faults you corrected?
By applying the path testing method, we nd the test cases in Table 11-1. The test cases for getNumDaysInMonth()
are the same as before since we did not change its ow of control. The test cases for isLeapYear(), however, are
public class MonthOutOfBounds extends Exception {};
public class YearOutOfBounds extends Exception {};
public class MyGregorianCalendar {
public static boolean isLeapYear(int year) {
boolean leap;
if ((year%400) == 0) {
leap = true;
} else if ((year%100) == 0) {
leap = false;
} else if ((year%4) == 0) {
leap = true;
} else {
leap = false;
}
return leap;
}
public static int getNumDaysInMonth(int month, int year)
throws MonthOutOfBounds, YearOutOfBounds {
int numDays;
if (year < 1) {
throw new YearOutOfBounds(year);
}
if (month == 1 || month == 3 || month == 5 || month == 7 ||
month == 10 || month == 12) {
numDays = 31;
} else if (month == 4 || month == 6 || month == 9 || month == 11) {
numDays = 30;
} else if (month == 2) {
if (isLeapYear(year)) {
numDays = 29;
} else {
numDays = 28;
}
} else {
throw new MonthOutOfBounds(month);
}
return numDays;
}
...
}
Figure 11-1 Corrected implementation of getNumDaysInMonth() and isLeapYear() methods (Java).
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
different, since we added two more decision points. Note that the additional test cases applied to the incorrect version
of isLeapYear() would have exercised the bugs.
112 Generate equivalent Java code for the statechart diagram for the Set Ti me use case of 2Bwatch (Figure 11-14).
Use equivalence testing, boundary testing, and path testing to generate test cases for the code you have just
generated. How do these test cases compare with those generated using state-based testing?
State-based testing is similar to applying equivalence testing for every state, hence, equivalence testing applied to a
state-driven system results in the same set of test cases. Path testing and boundary testing do not generate any new
tests.
113 Build the statechart diagram corresponding to the Pur chaseTi cket use case of Figure 11-24. Generate test
cases based on the statechart diagram using the state-based testing technique. Discuss the number of test cases
and differences with the test case of Figure 11-25.
Figure 11-2 depicts a possible statechart diagram and a series of derived test cases for this exercise. The main
difference between these test cases and those in the book is that they have been more systematically generated, and
thus, cover more possible transitions. It should also be noted, however, that this set of test cases depends on the level
of detail statechart diagrams. For example, in this sample solution, we did not differentiate between the case when the
user inserts the exact amount and the case when the user inserts more money than necessary. Doing so would have
added an additional state and transitions, and consequently, additional test cases. Moreover, this technique fails to
detect omissions error as in the case of path testing the isLeapYear() method. If behavior is missing from the system
or the statechart, it is unlikely that a test case would uncover such a defect.
Table 11-1 Test cases and their corresponding path for the corrected implementation of getNumDaysInMonth() an isLeapYear()
methods.
Test case Path
For getNumDaysInMonth()
(year = 0, month = 1) {throw1}
(year = 1901, month = 1) {n=31 return}
(year = 1901, month = 2) {n=28 return}
(year = 1904, month = 2) {n=29 return}
(year = 1901, month = 4) {n=30 return}
(year = 1901, month = 0) {throw2}
For isLeapYear()
(year = 2000, month = 1) {leap = true return}
(year = 1900, month = 1) {leap = false return}
(year = 1904, month = 1) {leap = true return}
( year = 1901, mont h = 1) { l eap = f al se r et ur n}
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Stimuli Transition tested Predicted resulting state
Empty set a. Initial transition
Idle
button 1 b.
Zone 1
button 2 f.
Zone 2
button 1 i.
Zone 1
button 3 g.
Zone 3
button 1 l.
Zone 1
button 2 f. Put the system into the Zone 2 state to
test the next transition.
Zone 2
button 3 j.
Zone 3
button 2 m. Zone 2
insert money < zone 2 amount h. Money Inserted
button 2 p. Zone 2
button 1 i. Put the system into the Zone 1 state to
test the next transition.
Zone 1
insert money < zone 1 amount e. Money Inserted
button 1 o. Zone 1
button 3 g. Put the system into the Zone 3 state to
test the next transition.
Zone 3
insert money < zone 3 amount k. Money Inserted
button 3 q. Zone 3
insert money > zone 3 amount k., n, r, s. Idle
button 2 b. Zone 1
insert money > zone 2 amount k, n, r, s. Idle
button 3 c. Zone 3
Figure 11-2 UML statechart diagram and resulting tests for the PurchaseTicket use case.
Idle
a. Initial transition
Zone 1
Zone 2
Zone 3
Money Inserted
Print Ticket Return Change
b.1
c.2
d.3
i.1
l.1
f.2
p.2
o.1
m.2
g.3
j.3 q.3
e.money
k.money
h.money
n.enough money
r.ticket printed
s.change returned
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
114 Given the following subsystem decomposition:
comment on the testing plan used by the project manager:
What decisions were made? Why? What are the advantages and disadvantages of this particular test plan?
This test plan was generated according to a sandwich testing plan, without unit testing subsystems from the target
layer and without double tests. The advantage is that this test plan consists of fewer tasks. The disadvantage is that
failures discovered during the Test A,B,C,D activity will be difcult to locate. A better strategy is to unit test the
subsystems in the target layer.
115 You are responsible for the integration testing of a system that encrypts network trafc. This system includes a
key generator subsystem that uses random numbers. During integration testing, you use a stub implementation
of the key generator that produces a predictable result. However, for the release version of the system, you want
to substitute the stub implementation with the random implementation, so that the generated keys are not
predictable to an outsider. Implement a test infrastructure using one of the design patterns described in
Chapter 8, Object Design: Reusing Pattern Solutions to enable the exchange of these two key generator
implementations at run time. Justify your choice.
We use a bridge pattern to substitute different implementations of the random number generator at run time.
116 Use path testing to generate test cases for all the methods of the Net wor kConnect i on class depicted in Figure
11-15 and in Figure 8-11. Expand rst the source code to remove any polymorphism. How many test cases did
you generate using path testing? How many test cases would you generate when the source code is not
expanded?
The receive() and setNetworkInterface() methods would yield only one test case without expanding the source
code, since they do not have decision points. However, receive() has one polymorphic message send that could be
bound to three different classes, resulting in three test cases. setNetworkInterface() has two that could be each
independently bound to three different classes, resulting in six test cases. See Table 11-2.
User Interface (A)
Billing (B) Event Service (C) Learning (D)
Database (E) Network (F) Neural Network (G)
Layer III
Layer II
Layer I
Test A
Test G
Test B,E,F
Test D,G
Test A,B,C,D
E,F,G
Test A,B,C,D,
Test E
Test F
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
117 Apply the software engineering and testing terminology from this chapter to the following terms used in
Feynmans article mentioned in the introduction:
What is a crack? [error]
What is crack initiation? [error introduction]
What is high engine reliability? [high component reliability]
What is a design aim? [design goal or a nonfunctional requirement]
What is a mission equivalent? [the operation of the engine for 500 continuous seconds, either during a
mission or during a test; this term has no equivalent in the testing chapter]
What does 10 percent of the original specication mean? [the achieved lifetime of the engine is a tenth
of the original goal, that is, instead of having a lifetime of 55 mission equivalents, it has a life time of 5.5
mission equivalents]
How is Feynman using the term verication, when he says that As deciencies and design errors are
noted they are corrected and veried with further testing? [In this context, verication means the
accumulation of sufcient experimental evidence to convince the designers that an error has been
corrected; in software engineering, verication is the development of a proof using formal methods to
achieve the same goal]
Figure 11-3 Sample solution for Exercise 115.
Table 11-2 Number of test cases produced with path testing, with and without source code expansion in Exercise 116.
Method
Number of test cases without
expanding source code
Number of test cases after
expanding the source code
send() 2 6
receive() 1 3
setNetworkInterface 1 6
GeneratorTestStub
RandomNumber Gener at or EncryptionAlgorithm
Client
ProductionGenerator
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
12. Rationale Management: Solutions
12-1 In Section 12.3, we examined an issue related to access control and notication in the CTC system. Select a
similar issue that could occur in the development and CTC and populate it with relevant proposals, criteria,
arguments, and justify a resolution. Examples of such issues include:
This exercise is best done in groups of three or more. The goal is not to test the students knowledge of the CTC
systems and its related issues, but rather, to test their ability to structure the debate resulting in this knowledge using
rationale techniques.
How can consistency between mainServer and hotBackup be maintained?
I[1]: Maintaining consistency between mainServer and hotBackup.
P[1]: The mainServer receives all the requests and forwards them to the hotBackup server after
they have been processed. Replies from hotBackup are simply ignored.
A[1] against P[1]: If the mainServer crashes after successfully precessing a request but
before forwarding it to the hotBackup, the states of both servers will be inconsistent.
P[2]: Requests to the mainServer are concurrently forwarded to the hotBackup (e.g., by a proxy
server standing between the clients and the servers). The replies from both servers are
received by the proxy but those from hotBackup are simply ignored.
A[2] for P[2]: This solution can ensure consistency if requests are processed sequentially
by both servers and if no processing occurs between requests.
A[3] against P[2]: This adds a component in the architecture (the proxy) which represents
a single point of failure.
P[3]: Requests and replies from the main server are rst logged and then forwarded to the
mainServer or the requesting client, respectively. When switching servers, the hotBackup
examines the log of requests to determine the state of the server at the crash time.
A[4] against P[3]: This solution introduces a delay during switching. Given a reliable
log, restarting the mainServer may be simpler.
R[1]: P[2] since P[1] does not solve the problem and P[3] is inefcient.
How should failure of the mainServer be detected and the subsequent switch to the hotBackup be
implemented?
I[2]: How to implement hot switching of servers?
P[2]: Requests to the mainServer are concurrently forwarded to the hotBackup (e.g., by a proxy
server standing between the clients and the servers). The replies from both servers are
received by the proxy but those from hotBackup are simply ignored.(same as P[2] above).
If the main server fails to respond within a time out, it is restarted and the hotBackup
becomes the mainServer.
A[5]: Simplicity.
P[4]: A similar proxy server forwards requests to both servers and forwards the rst reply that
comes back to the client. If the second reply does not occur within a time out, the slower
server is restarted.
A[6] for P[4]: Better response time.
A[7] against P[4]: To take advantage of the shorter response time, the proxy has to be
multithreaded, adding complexity to a single point of failure.
R[2]: P[4] for response time.
122 You are developing a CASE tool using UML as its primary notation. You are considering the integration of
rationale into the tool. Describe how a developer could attach issues to different model elements. Draw a class
diagram of the issue model and its association to model elements.
This is an open ended question. The main point that should be present in all correct answers is that, ideally,
developers should be able to attach any type of rationale node to any type of modeling element. For example,
developers could attach issues on individual methods, develop proposals which refer to proposed additional methods
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
and attributes, and discuss arguments which refer to existing model elements. The creation of two abstract classes,
ModelElement and RationaleNode, would make this type of functionality easier to implement (see Figure 12-1).
123 Below is an excerpt from a system design document for an accident management system. It is a natural
language description of the rationale for a relational database for permanent storage. Model this rationale
with issues, proposals, arguments, criteria, and resolutions, as dened in Section 12.3.
This exercise tests the students knowledge of issue models. The student can either draw a graph or use idented text to
represent the resulting issue model. The text below shows a simple representation using idented text and tags. Once
the issue model is completed, the reader may notice that some of the proposals were not thoroughly discussed and
argued and that one of the pseudo requirements (modeled as a criterion) was not met.
Key: I[1] = issue 1, P[1] = proposal 1, A[1] = argument 1, C[1] = criterion 1, R[1] = resolution to I[1].
I[1]: Which database engine realization?
P[1]: OO DBMS.
A[1] for P[1]: Can handle complex data relationships.
A[2] for P[1]: Buzzword compliant. (see C[1])
A[3] against P[1]: Sluggish for large volumes of data or high-frequency accesses.
A[4] against P[1]: Existing products do not integrate well with CORBA. (see C[2])
P[2]: Relational DBMS.
Figure 12-1 Example class diagram for exercise 123 (some classes omitted for clarity)
One fundamental issue in database design was database engine realization. The initial
nonfunctional requirements on the database subsystem insisted on the use of an object-oriented
database for the underlying engine. Other possible options included using a relational database, a
le system, or a combination of the other options. An object-oriented database has the advantages
of being able to handle complex data relationships and is fully buzzword compliant. On the other
hand, OO databases may be too sluggish for large volumes of data or high-frequency accesses.
Furthermore, existing products do not integrate well with CORBA, because that protocol does
not support specic programming language features such as Java associations. Using a relational
database offers a more robust engine with higher performance characteristics and a large pool of
experience and tools to draw on. Furthermore, the relational data model integrates nicely with
CORBA. On the downside, this model does not easily support complex data relationships. The
third option was proposed to handle specic types of data that are written once and read
infrequently. This type of data (including sensor readings and control outputs) has few
relationships with little complexity and must be archived for extended periods of time. Files offer
an easy archival solution and can handle large amounts of data. Conversely, any code would need
to be written from scratch, including serialization of access. We decided to use only a relational
database, based on the requirement to use CORBA and in light of the relative simplicity of the
relationships between the systems persistent data.
ModelElement
Class
Attribute
Association
Operation
RationaleNode
Issue
Proposal
Resolution
Argument
* *
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
A[5] for P[2]: More robust engine than OO DBMS.
A[6] for P[2]: Higher performance characteristics.
A[7] for P[2]: Large pool of experience.
A[8] for P[2]: Some relational DBMS integrate well with CORBA.(see C[2])
P[3]: File system
A[9] for P[3]: Good for serial data with simple relationships, including sensor data.
A[10] against P[3]: Need to rewrite many programs to provide similar functionality than a
DBMS, including serialization of access.
P[4]: Combination of P[1,2,3].
C[1]: Database subsystem should use an OO database.(Pseudorequirement)
C[2]: Needs to co-exist with CORBA. (Prior design decision)
R[1]: P[2], because of C[2] and simplicity of relationships of systems persistent data.
124 Consider the NFR Framework described in Section 12.3.7. Draw a QOC model that is equivalent to the goal
graph depicted in Figure 12-12. Discuss the respective advantage and disadvantages of QOC and the NFR
Framework for representing rationale during requirements.
Figure 12-2 depicts a QOC model equivalent to the goal graph. The advantages of the NFR Framework over QOC are
that
goals (questions) are represented as an AND-OR renement a hierarchy
the satised goals for a given option are more easily visible
125 You are integrating a bug reporting system with a conguration management tool to track bug reports, bug
xes, feature requests, and enhancements. You are considering an issue model for integrating these tools. Draw
a class diagram of the issue model, the corresponding discussion, conguration management, and bug-
reporting elements.
This exercise can be answered only after students have read Chapter 13, Conguration Management. This exercise is
with Chapter 12, however, since it is relevant to rationale management.
This exercise is also an open-ended question, as exercise 122 The solutions of this exercise, however, should
different signicantly from those of exercise 122, as the integration problem is only supercially related.
Problem reporting and conguration management tools are usually based on the following workow: a change is
requested, the change is discussed, assessed and planned, and the change is realized, test, and incorporated into a
future release. In this exercise, change requests can be feature requests or bug reports. During the assessment of these
proposed changes, developers raise issues, propose alternative solutions, and eventually come to a consensus in the
form of a resolution. This resolution is then realized as a change.
Figure 12-2 QOC model for Exercise 124.
Acccount+PIN
FingerPrintReader
Flexibility
SmartCart+PIN
Authentication?

Low Cost
:Option
:Option
:Option
:Question
:Criterion
:Criterion
+
+
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
In the class diagram representing these concepts (Figure 12-2), the main associations between both models are
between Change Request and Issue, and between Resolution and Change. Note that both associations are many to
many.
Figure 12-3 Example class diagram for exercise 123 (some classes omitted for clarity)
Bug Report
Issue
Proposal
Resolution
Argument
Feature Request
Change Request
Bug Fix Enhancement
Change
* *
* *
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
13. Conguration Management: Solutions
13-1 RCS adopts a reverse delta approach for storing multiple versions of a le. For example, assume a le has three
revisions, 1.1, 1.2, and 1.3, RCS stores the le as of version 1.3, then, the differences between 1.2 and 1.3, and
the differences between 1.1 and 1.2. When a new version is created, say 1.4, the difference between 1.3 and 1.4
is computed and stored, and the 1.3 version is deleted and replaced by 1.4.
Explain why RCS does not simply store the initial version (in this case 1.1) and the differences between each
successive version.
The trade-off between reverse deltas and forward deltas is a response time trade-off. In the case of reverse deltas (as
RCS does), the time to retrieve the latest version is the shortest, while time to retrieve the oldest versions is the
longest. In the case of forward deltas, the time to retrieve the initial (oldest) version is the shortest. In most projects,
retrieving the latest version is the most common case. Note that storing reverse deltas increases the time for creating a
new version.
132 CVS uses a simple text-based rule to identify overlaps during a merge: There is an overlap if the same line was
changed in both versions that are being merged. If no such line exists, then CVS decides there is no conict and
the versions are merged automatically. For example, assume a le contains a class with three methods, a(),
b(), and c(). Two developers work independently on the le. If they both modify the same lines of code, say the
rst line of method a(), then CVS decides there is a conict. Explain why this approach will fail to detect
certain types of conict. Provide an example in your answer.
This approach fails when developers introduce conicts when modifying different lines of a le. For example, if two
developers introduce a new eld with the same name in the same class in different locations in the le. Note that
developers can also introduce semantic conicts by modifying two different les (e.g., by making different
assumptions about the same class).
133 Conguration management systems such as RCS, CVS, and Perforce use le names and their paths to identify
conguration items. Explain why this feature prevents the conguration management of CM aggregates, even in
the presence of labels.
Note to the instructor: this is a difcult question for students who have not used conguration management systems,
in particular branch management.
The main issue with using path names for identifying conguration items is that path names are attributes of
conguration items and thus, can also change. Assume for example that directories represent subsystems (e.g.,
storage, notication, user interface, train tracking). In the case of a change of the subsystem decomposition, the
directory structure may change (e.g., the storage directory may be split into a database directory and a JDBC
directory). For tools which use path names as identiers, this translates into conguration items being deleted and
added under a different path name. Since different versions of the same conguration items have different identiers,
operations such as merging branches are consequently not possible.
134 Explain how conguration management can be benecial to developers, even in the absence of a change
control or auditing process. List two scenarios illustrating your explanation.
Scenario 1: Assume developers work on new features sequentially and create new versions of the systems as a feature
becomes stable. Assume the current version is n and that versions m through n contain partial implementation of the
latest feature F. The implementation of F unfortunately, introduces many issues and developers decide to rethink their
approach. They nd a better way to implement F, however, it requires them to undo all changes related to F. When
using a version control tool and the sequential conguration management policy described before, this can be simply
done by checking out version m-1 and discarding versions m through n.
Scenario 2: Branches can be used to manage the development of concurrent features. If the implementation of all
features succeed, the branches need to be merged. If the development of one feature fails as in scenario 1, the
corresponding branch can be ignored while the rest of the development proceeds.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
135 In Chapter 12, Rationale Management, we described how rationale information can be represented using an
issue model. Draw a UML class diagram for a problem tracking system that uses an issue model for the
description and discussion of changes and their relationship with versions. Focus only on the domain objects of
the system.
Clarication: domain objects here means entity objects.
This exercise has many different possible solutions since it is a design exercise. Figure 13-1 depicts a possible
solution. The key element is that the release of a new version triggers the discovery of new problems, which are then
discussed (cause identied, solution proposed and evaluated, and decision made). The decision then results in a new
version aimed at correcting the new problem.
136 In Chapter 11, Testing, we described how the quality control team nd faults in promotions created by
subsystem teams. Draw a UML activity diagram including the change process activities and testing activities of
a multiteam project.
Figure 13-10 in the book is a sample solution to this problem. The key elements the instructor should look in student
solutions are:
Testing is performed by the quality control team.
The quality control team sends change requests or problem reports to the subsystem teams.
Changes are performed by the subsystem teams.
Subsystem teams send promotions to the quality control team.
The subsystem teams decide when to create promotions.
The quality control team decides when to create a release.
Figure 13-1 Example solution for exercise 10.5.
Alternative
Decision
Criterion
Issue
assesses
solvableBy
resolvedBy
based-on
* * *
implementedBy
RationaleSubsystem
CMSubsystem
Controlled item
*
Version CM Aggregate
*
Configuration item
raised by
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
14. Project Management: Solutions
14-1 In Figure 14-1, we modeled the phases of a project with a state chart. Use the state pattern by making each of
these phases a different class. Use the project management activities introduced in this chapter to make them
public operations on these classes. Assume a team-based project organization.
Figure 14-1 depicts a sample solution to this exercise. The Pr oj ect Phase class denes operation for each project
management activity class. Each concrete state class (one per phase) denes operations for those operations that it
can handle. All other operations are implemented so that they trigger a change of state. The context is stored in the
Project class.
142 What are the relative advantages of at stafng versus gradual stafng?
The advantage of gradual stafng is that saves resources in the early part of the project. The advantage of at stafng
is that all the ramp up effort (e.g., training, team formation, denition of procedures, etc.) is moved up front.
143 What is the difference between status meetings and decision meetings? Should they always be kept separate in a
meeting?
The objective of a status meeting is for team members to report status to the team leaders, in an honest, veriable, and
unbiased manner. This enables the team leader to track progress and plan corrective decisions if necessary, such as
adding resources or creating new communication paths.
The objective of a decision meeting is either, for a team leader to communicate a decision to the team and ensure it is
implemented through action items, or for a team to discuss a set of issues and achieve a consensus.
It is better to keep status and decision separate, such that team members do not feel punished for reporting problems.
A team leader needs the team members to report problems as soon as possible and can encourage this if the team
leader gives the opportunity for the team to resolve the problem rst by themselves. If the problem worsen, then the
team leader may decide to intervene or to call a decision meeting in which the team cooperatively designs a solution.
In any case, there should be a time interval between the reporting of a problem and its corresponding intervention.
144 Why should the role of architect and team leader be assigned to distinct persons?
Figure 14-1 Sample solution for Exercise 14-1.
Project Pr oj ect Phase
f or mul at eI dea( )
doCost Benef i t Anal ysi s( )
doFeasi bi l i t ySt udy
doRevi ew
doPr obl emSt at ement
. . .
Conception
formulateIdea()
doCostBenefitAnalysis()
doFeasibilityStudy
doReview
doProblemStatement
...
Definition
formulateIdea()
doCostBenefitAnalysis()
doFeasibilityStudy
doReview
doProblemStatement
...
Termination
formulateIdea()
doCostBenefitAnalysis()
doFeasibilityStudy
doReview
doProblemStatement
...
. . .
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
The role of architect requires good technical abilities. The role of team leader requires good management abilities.
Both kinds of abilities are rarely present in the same person. Both roles are time consuming can rarely be
accomplished well by the same person at once.
145 Draw a UML model of the team organization of the OWL project for each the three main phases (i.e., initiation,
steady state, termination).
An example solution to this exercise is depicted in Figure 14-1. During project initiation, the management team is
responsible for the planning and other management activities, while the requirements team denes the system with
the users and the client and the architecture team design a preliminary architecture. During steady state, a team for
each subsystem is created, along with a documentation team and a quality control team. During project termination,
the subsystem teams are replaced with a single installation team which is responsible for deploying the system and
xing any last minute problems discovered by the quality control team.
146 Draw a task model of the system design of the MyTrip system presented in Chapter 6, System Design:
Decomposing the System.
A task plan can be represented as an activity diagram. The MyTrip system was used as an example to illustrate each
system design activity. Hence, the activity diagram of Figure 6-28 can be used as a starting point for the solution to
this exercise. We remove the Implement Subsystem activity (it is not part of system design) and add a Rene
subsystems activity to plan for a second iteration (see Figure 14-2). A conservative manager may even want to
serialize all tasks instead of conducting the activities of system design in parallel.
Figure 14-2 Example solution for exercise 11.4 (UML class diagram).
Management
:Team
Architecture
:Team
Architecture
:Team
Management
:Team
UserInterface
:Team
FacilitiesMgt
:Team
Control
:Team
DatabaseMgt
:Team
Visualization
:Team
QualityControl
:Team
Documentation
:Team
Management
:Team
Documentation
:Team
QualityControl
:Team
Installation
:Team
Project initiation
Steady state
Project termination
Requirements
:Team
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
147 Estimate the time to complete each task of the task model produced in Exercise 146 and determine the critical
path.
Note to the instructor: the estimates provided by the students need not (and will not) be realistic. The goal of the
exercise is to check the understanding of the concepts of critical path and task dependency. The example solution in
Figure 14-3 depicts the task model as a PERT chart and indicates the critical path in bold.
Figure 14-3 Example solution for exercise 11.5 (UML activity diagram).
Figure 14-4 Example solution for exercise 11.6 (PERT chart).
Describe boundary
conditions
Define Define
subsystems
Map subsystems
to hardware/
Manage
Select a
Define access
design goals
persistent data
control policies
global
software platform
control flow
Refine
subsystems
Define design
goals
1
Jan 9
1d
Jan 10
Map subsystems to
hw/sw platform
3
Jan 16
1d
Jan 16
Manage persistent
data
4
Jan 16
2d
Jan 17
Define access
control policies
4
Jan 16
2d
Jan 17
Select global
control flow
4
Jan 16
1d
Jan 16
Describe boundary
conditions
4
Jan 16
5d
Jan 20
Define subsystems
2
Jan 10
4d
Jan 13
Refine subsystems
2
Jan 23
5d
Jan 27
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
148 Identify, prioritize, and plan for the top ve risks related to the system design of MyTrip presented in Chapter 6,
System Design: Decomposing the System.
Note to the instructor: Again, the risks provided by the students need not be realistic. They do need, however, to be
risks. The goal of the exercise is to make the students think about what could go wrong and to check their
understanding of the risk concept. Table 14-1 is an example solution of risks and corresponding mitigation strategies.
The risks are ordered by priority (most important rst).
149 Identify, prioritize, and plan for the top ve risks related to the user interface subsystem of CTC presented in
Chapter 12, Rationale Management.
Same exercise as before with a different application domain. Table 14-1 depicts an example solution.
Table 14-1 MyTrip risks
Risk Mitigation strategy
MyTrip presents a safety hazards for drivers. Design a user interface that only allows passengers to
consult trips.
Describe the risks of using MyTrip while driving on the
rst page of the user manual.
MyTrip is not secure. Plan for security tests by the quality control team.
As part of maintenance, plan for a security team to
continuously test and address security vulnerabilities.
MyTrip is not reliable. Underestimate reliability of components, in particular,
the hardware components over which the project has no
control: phone network, home computer, etc.
Plan for the release of patches and upgrades to address
serious bugs.
A competitor develops conicting standards for storing and
planning trips
Design for modiability.
The number of users and trips were underestimated. MyTrip
does not scale beyond requirements
Design for more users and trips than required.
Table 14-1 CTC user interface risks
Risk Mitigation strategy
The user interface does not meet the availability criteria. Plan for extensive eld tests before deployment.
Plan for an emergency maintenance team to address any
serious problems.
Plan to roll back to the older system if the availability
criteria is not met.
The user interface consumes too many system resources. Plan for early performance testing.
Select hardware that can accommodate more load than
initially required.
The user interface does not meet the usability criteria Plan for usability tests with a representative set of users.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
1410Linux, developed using the bazaar model, is more reliable and more responsive than many operating systems
running on Intel PCs. Discuss in some detail why the bazaar model should, however, not be used for the Space
Shuttle control software.
One of the key of the bazaar model is that many users are actually developers willing to diagnose problems and
propose solutions. The system, once released, goes through a large scale and parallel test which will nd most of the
important problems. The Space Shuttle does not have many users. Moreover, the price of discovering a bug in the
eld is much more serious.
1411Organize the project participants into teams of four people. Each team has the following resources available: 2
eggs, 1 role of TESA lm, 1 role of toilet paper, a cup with water, a bucket with two liters of sand, 20 foam balls
(each about 1 cm diameter), 1 table whose surface is about 1 meter above the ground. Each team has 25
minutes time to build and test an artifact that allows an egg to be released 75 cm above the table such that the
egg falls on the table without cracking. Each team has another 5 minutes to demonstrate the artifact to project
management.
This is a team building exercise that forces groups of individuals who do not know each other to work together,
identify each individuals strengths, and recognize that some problems are better solved in a collaborative
environment. Such exercises are done at the beginning of a project to form teams.
1412Organize the project participants into teams of four people. Each team has the following resources available: 2
buckets of DUPLO pieces, 2 tables separated by a distance of 1.50 meters. Each team has 25 minutes time to
build and test a bridge solely out of the available DUPLO pieces that hangs free for at least one minute
between the two table surfaces. Each team has another 5 minutes to demonstrate the artifact to project
management.
This is another team building exercise like Exercise 1411.
1413Specify all management models for the icebreaker project described in Exercise 14-11.
After the completion of Exercise 1411, the instructor should select a team that displayed a minimum of structure
during the problem solving (as opposed to a single participant coming up with the solution) and ask participants
reverse engineer a schedule, roles, teams, and a task model. This should illustrate both the usefulness of such models
to reason about the team dynamic and their limitations.
1414Write an SPMP for the icebreaker project described in Exercise 14-11.
This exercise has the same purpose as Exercise 1413.
The user interface represents a paradigm shift (i.e., the users
might be confused by the new interface)
Ensure that users have sufcient training before they
dispatch trains with the new system.
The user interface does not scale to a larger number of trains
and dispatchers
Design for a larger number of trains and dispatchers than
initially required.
Table 14-1 CTC user interface risks
Risk Mitigation strategy
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
15. Software Life Cycle: Solutions
15-1 Model as classes the activities of Figure 15-2 and the work products of Figure 15-4 and draw a UML class
diagrams depicting the relationships between activities and work product.
Figure 15-5 in the book is a possible solution.
152 Assume that the waterfall model shown in Figure 15-8 has been derived from the IEEE standard model in
Figure 15-7 during the activity Life Cycle Modeling. What processes and activities have been omitted in the
waterfall model?
The goal of this exercise is to have students think about processes which are needed all the time, e.g., conguration
management. The following processes were omitted from the waterfall model:
Project management processes: Project Monitoring & Control, Software Quality Management,
Integral processes: Conguration Management, Documentation Development, Training,
Post development processes: Maintenance, Retirement.
The project management and integral processes that were omitted run for the complete duration of the project, and
thus, do not t in the sequential model of the waterfall. The Maintenance and Retirement processes could simply
added at the bottom of the water fall. They were probably omitted because those activities are not considered part of
development.
153 Redraw Boehms spiral model in Figure 15-10 as a UML activity diagram. Compare the readability of the
original gure with the activity diagram.
Figure 15-1 depicts an example solution for this exercise. Without the spiral metaphor, it is difcult to understand that
the life cycle is incremental and iterative.
154 Draw a UML activity diagram describing the dependency between activities for a life cycle in which
requirements, design, implementation, test, and maintenance occur concurrently. (This is called an
evolutionary life cycle.)
Figure 15-2 depicts an example solution for this exercise.
Figure 15-1 Activity diagram for exercise 12.5 (only first three cycles shown)
Design
Validation
Requirements
Validation
Concept of
Operation
Requirements
Plan
Requirements
Specification
Development
Plan
System
Design
Risk
Analysis
Risk
Analysis
Integration
Plan
Integration
Risk
Analysis
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
155 Describe how testing activities can be initiated well before implementation activities. Explain why this is
desirable.
Testing activities can be initiated as soon as requirements activities are completed. System tests are generated from
the requirements. Integration tests are generated from the system design. Unit tests are generated from the object
design. Initiating testing activities early has two main advantages:
Some types of faults can be found merely by developing test cases, before executing them (e.g., interface issues).
The test cases will not be biased by the implementation, and thus focus on the requirements of the system.
156 In project management, a relationship between two tasks is usually interpreted as a precedence relationship;
that is, one task must complete before the next one is initiated. In a software life cycle, a relationship between
two activities is a dependency; that is, one activity uses a work product from another activity. Discuss this
difference. Provide an example in the context of the V-model.
A precedence relationship indicates that a task must end before the succeeding task can start. A dependency
relationship, however, does not always imply that the dependent activities must proceed in sequence. In the V-model,
for example, the component integration and test activity depends on both the preliminary design activity (for the
subsystem decomposition) and the unit test activity (for the unit tested system). The component integration and test
activity can start with the planning of the integration tests as soon as the preliminary design is complete, which can
occur even before the unit test activity is initiated.
157 Assume you are part of the IEEE committee that will revise the IEEE 1074 standard. You have been assigned
the task of modeling communication as an explicit integral process. Make a case for the activities that would
belong to this process.
Since communication would be an integral process, all activities in this process group would be process independent
and phrased in similar terms as the conguration management or documentation activities. Examples would include:
Plan communication infrastructure
Deploy communication infrastructure
Plan reviews
Dene change request procedures
Plan issue tracking
etc.
Figure 15-2 Activity diagram for exercise 12.6
Problem
Definition
Requirements
Design
Implementation
Test
Maintenance
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
16. Methodologies: Putting It All Together
16-1 What is the difference between a software life cycle and a methodology?
A software life cycle is a model of the activities and work products necessary to develop a system. A methodology is
a collection of methods and tools for developing a system and species when methods or tools should be used and
what to do if unexpected events occur. Methodologies and life cycles are complementary: a software life cycle can be
realized using a variety of software engineering methodologies. Although specic methodologies may imply a
specic life cycle, in general, a methodology can be applied in a different life cycles.
Examples of life cycles: Boehms spiral model, the Unied Process, the Waterfall model.
Examples of methodologies: Royces unied framework, Extreme Programming.
162 You are improving a legacy system. You need to customize a software life cycle for your project. Which activities
would you need? In which order?
Any answer to this question should include a life cycle with an inventory analysis activity (to assess what work
products exist), one or more reverse engineering activities (to recreate missing work products), and change activities
to change or improve the aspect of the legacy system under consideration.
163 Royce uses a management metric that counts the number of participants who leave the project or join the
project. You are managing a multi-team project and notice that the staff turnover in one team is high.
Hypothesize what causes could lead to such a symptom. For each cause, propose a remedy.
High-staff turnover is a negative indicator. It points to a fundamental problem that can only worsen if not addressed.
Some of the cause and remedies are depicted in Table 16-1
164 The heuristics we outlined in Section 16.5.4 indicate that the need for models is higher in distributed
organizations. Open-source projects are highly distributed projects that follow an entity-based life cycle and
typically do not have requirements or system design documents. Provide examples of how modeling knowledge
is made explicit and transferred among participants in such cases.
Open source projects typically feature models that can be easily rened or extended incrementally. Examples include
information stored in change tracking logs, conguration management logs, frequently asked questions, and how-tos.
165 Royces methodology considers six project factors (scale, stakeholder cohesion, process exibility, process
maturity, architectural risk, and domain experience) when tailoring a process for a specic project
(Section 16.4.1). Use these factors to describe which types of projects could use XP as an appropriate
methodology. Justify your choice for each factor.
Table 16-1
Cause Remedy
Unclear or unrealistic goals Provide better tools to meet the goals or redene goals.
Unrealistic schedule Revisit schedule or increase resources.
Conict with team leader Replace team leader.
Skills needed for the team are better rewarded in other
organizations.
Revise reward structure or determine if required skills are
needed.
Object-Oriented Software Engineering: Using UML, Patterns, and Java: Solutions to Exercises
2004 Prentice Hall November 14, 2003
Table 16-2 discusses the types of projects for each factor.
166 While working with Staphylococcus bacteria in 1928, Sir Alexander Fleming accidently dropped some bread
crumbs on one of the dishes. After a week the bacteria on this dish did not grow as expected. Fleming noticed a
bacteria-free circle around a mold that was contaminating the staphylococcus culture. Instead of throwing the
dish away because the planned experiment had failed, he isolated the mold, grew it in a uid medium, and
discovered a substance that prevented growth of the bacteria even when it was diluted 800 times. Discuss the
discovery of penicillin using the terminology and issues introduced in this chapter.
This is an example of redening project goals. The initial experiment of Fleming failed, as he contaminated the petri
dish. However, Fleming recognized that he discovered a substance that was much more signicant than the original
experiment.
167 Based on the assumption that the earth was round, Columbuss goal was to nd a shorter way to India by going
west instead of east. He encountered America instead of India. Discuss the discovery of America by Columbus
using the terminology and issues introduced in this chapter.
Columbus and his contemporaries thought he succeeded in discovering a new route to India. From his perspective, he
succeeded. Later, after maritime routes around Africa were discovered and when Magellans crew circumvented the
globe, it became clear that Columbus had found a new continent. In the terminology introduced in Chapter 1, this is
also an example of how a theory is devised, validated, and falsied, as knowledge is accumulated and tested against
the theory.
168 Select a project you have been involved with. Categorizing into methodological issues as dened in this chapter,
describe the methodological trade-offs that occurred in the project.
The discussion should include a description of the project environment followed by a discussion of each
methodological issue, including the options that were considered and those that were selected.
Table 16-2
Factor XP value
Scale Low to medium. The maximum number of participants that XP can
accommodate is limited to the number of participants who can meet at the
same time during the daily stand-up meeting.
Stakeholder cohesion High. As many requirements change during the project, it is critical that the
client and developers be able to agree informally on important decisions.
Process exibility High. XP has been designed for projects that feature a client who is willing
to be collocated with the project and to rene requirements and estimates at
the beginning each iteration.
Process maturity Medium. XP projects need a critical mass of experienced participants to be
effective. This knowledge is then transmitted to the novice participants
during pair programming and peer reviews. An insufcient number of
experienced developers knowledgeable in XP will result in difculty to
manage the expectations of the client.
Architectural risk High. XP can accommodate architectural changes until late in the project, as
the architecture is incrementally rened with the rest of the code.
Domain experience Low. A collocated client enables developers to learn domain knowledge
rapidly.