Applying the object-oriented paradigm to the development of software requires individuals and teams to think and act differently than when designing procedural projects. While proponents of the object paradigm often say identifying objects is a simple and intuitive process, experienced developers know that this is not always true. The solution is the CRC (Classes, Responsibilities, Collaboration) Card method, a proven technique for identifying classes and visualizing and testing different class-based models during the design phase.
The CRC Card Book covers the CRC method from start to finish, illustrating its application in three different in-depth case studies which supply tips and pointers drawn from real world experience. The case studies are presented in the engaging style of a novella to demonstrate how personalities and organizational culture come into play when using the CRC technique. C++, Java, and Smalltalk experts provide implementation examples in each language. The CRC Card Book demonstrates how to:
Includes language implementation examples from such noted experts as:
About the Authors.
1. CRC: A Problem-Solving Technique.
The CRC Card.
CRC Is Fun!.
Problem Solving and Insight.
A Parable of Insight.
CRC Cards as a Metacognitive Process.
The Challenge of Finding Classes.
Making CRC Work: Project Management Guidelines.
Thinking About the World as Objects.
Instantiation: Classes and Objects.
Inheritance: Classes, Super-classes and Sub-classes.
Frameworks and Patterns.
A Case in Point.
Who Writes CRC Cards?
Two Is Only Company; Ten Is a Crowd.
Reading the CRC Card.
Brainstorming: A Discovery Technique.
Description of Brainstorming Principles.
Using Brainstorming to Find Classes.
The Candidate Class List: Where Do Classes Come From?
Read Over All Requirements Documents.
Look Carefully at Reports.
Examine Documentation and Files.
Identify Critical Classes.
Clarify System Scope.
Take Advantage of Hot Spots, Frameworks, and Patterns.
Phantoms and Ghosts, Surrogates and Aliases.
Distinguish Attributes from Classes.
The Annotated Candidate Class List.
Questions to Ask of a "Good Class".
Writing Classes on the CRC Cards.
"WHAT and Not HOW."
Responsibility for Behavior.
Responsibility for Knowledge.
Responsibility Detection Pointers.
Saver, Soone, & Eayrne's Cyber-Trading System.
Selecting the CRC Team.
Warming Up the Team.
Brainstorming Candidate Classes.
Selecting Core Classes.
The Theory Behind the Practice.
The Problem of Shared Meaning.
Beware the Limits of Working Alone.
Active Learning: The Rationale for Interactive Review.
How CRC Role Play Works.
Developing the Role-Play Scenarios.
The Theory of "Aha!"
The Psychological Model.
Confusion is Rarely Permanent.
Role Play: Performing the System Simulation.
Timing the CRC Role-Play Session.
Fashion Pro: A New Idea in Garment Creation.
The Fashion Pro Analysis Team.
Recapping the Core Classes.
Reusing Analysis Frameworks.
Continuing the Scenario List.
Object-Oriented Development Changes the SDLC.
The Waterfall SDLC.
The Spiral SDLC.
The Incremental and Iterative SDLC.
Using the OO SDLC to Manage Projects.
The Pilot Project.
Pilot Project Guidelines.
Choosing a Pilot from Our Case Studies.
Getting Started in Your Organization.
SWAT Teams or Gurus?
Accommodating Legacy Systems.
Legacy Integration Guidelines.
Communication by Executables.
Direct Reuse in Responsibilities.
Metrics for CRC Analysis.
Measuring the CRC Analysis.
The Next Step.
Traffic Problems in Chapel Hollow.
The CRC Card Team.
The Project Plan.
Segment 1: Simple Intersection.
Segment 2: Complex Intersection.
Segment 3: Zone Control.
Traffic Management Concepts.
Role Play: The First Enactment.
Assessment: Evaluating the Role Play.
Enactment Two: Verifying the Cards.
Using Role Play to Extend the System.
The Smalltalk Programming Environment.
Smalltalk Coding Guidelines.
Using CRC Cards with Smalltalk.
Analysis versus Program Design.
Going from Analysis to Design and Code.
First Test: Implementing the Detector.
Second Test: A Simple Intersection.
Third Test: A Full Intersection.
Fourth Test: The Specified Intersection.
Saving the Prototype.
Tips and Resources for Further Study.
The Code: Smalltalk Test Method Four.
Output from Smalltalk Test Four.
Developing C++ Systems.
Class Interface Design.
C++ Coding Guidelines.
Creating the Traffic System Class Interface.
Public Implementation of Constructors.
Private Implementations in the Traffic System.
List & Queue Services.
The Intersection Class.
The Phase Class.
C++ OO Design Priorities.
The Traffic System I/O Framework.
Moving to JAVA.
The Java Approach to the Traffic System.
Pointers for More.
CRC Cards Have Limits.
Why Methodologies Are Powerful.
Methodology Adoption Is a Must.
Once Over the Unified (Booch) Method.
Booch Domain Analysis.
Booch System Design.
Once Over the Shlaer-Mellor Method.
Summing Up Shlaer-Mellor.
Using Formal Methods for the Traffic Case.
Unified (Booch-Rumbaugh) Notation.
CRC Card Software.
Applying the object-oriented paradigm in the development of software applications requires the individual developer and the application team to think and act quite differently than one would in approaching a procedures based project. Object-oriented projects follow a new software development life cycle, one that is both iterative and incremental, a cyclic spiral of analysis, design, and deployment. In this new application-development process, determining and defining properly the classes that are central to the desired system at the beginning of the life cycle is critical. Thorough analysis of the problem and good design up front saves time, and money and helps ensure a successful end result.
Although proponents of the object paradigm often say that identifying objects is a simple and intuitive process, a number of noted experts admit that this is not always true! Particularly with larger-scale applications, omitting a formal analysis of the base classes necessary to the application and the related analysis of their responsibilities and collaboration is certain to lead to missed schedules, blown budgets, and frayed nerves. The solution is to use the CRC process to determine the classes necessary to the system as part of the design process for the application. CRC (classes, responsibility, and collaboration) cards can be used to visualize and test different class-based models during the design phase. It is a proven technique used and advocated by leading methodologists.
The CRC Card Book demonstrates the use of the CRC methodology in a realistic team setting, covering the full range this methodology from initial identification of classes to the production of code based on these classes. At this writing, there are two other books on or incorporating coverage of CRC. Rebecca Wirfs-Brock wrote Designing Object-Oriented Software many years ago. It remains the classic definition of responsibility-driven design, but does not discuss the application of the technique in the team-oriented setting in which large-scale applications are developed. Nancy Wilkinson's book, Using CRC Cards, focuses specifically on the C++ programming community and likewise does not include any illustration of the team approach to class discovery.
The approach of The CRC Card Book is to cover the CRC method from start to finish, demonstrating its application in three different, detailed case-study examples while supplying tips and pointers throughout. The book demonstrates how real teams can use the CRC technique to accomplish a variety of tasks, including:
The CRC Card Book is organized as follows:
We place special emphasis on the dynamics of team work, and how these dynamics are applied to the successful use of CRC. The first team strategy we suggest is brainstorming. We show how this applies to the task of finding classes and listing scenarios.
The second team strategy we recommend is role play. CRC cards provide a uniquely dynamic way of testing out your conception of the system and evaluating potential paths of collaboration. Role plays get everyone involved and invested in the system.
Three case studies are provided to help you visualize how the guidelines developed in the descriptive chapters might be applied in a real-life situation. All three case studies are based on real systems. They are presented in a dramatic, novella style to demonstrate more clearly how personalities and organizational culture come into play when a group is using the CRC technique. This is especially important in the context of brainstorming and role play. We hope that this approach not only will clarify the use of the method, but will provide a little entertainment along the way!
We also provide two examples of how our third case study, the control of automobile traffic intersections, might look when a programmer uses CRC cards as the basis for coding in Smalltalk, C++, and JAVA. Three experts in each language community joined with us to write this unique material.
Following the examination of role playing in the CRC method, we discuss the overall demands of managing an object-oriented project, along with suggestions as to useful metrics for monitoring the process, and for including legacy software in OO systems. We also provide some ideas for the transition from informal modeling with CRC cards to the use of a full-blown, formal methodology such as Schlaer-Mellor or the Unified Modeling Language. The final case study in the book provides additional insight into the application of these techniques.
In this concluding chapter we discuss the limitations of CRC cards and examine several popular and more comprehensive object-oriented analysis and design methodologies.
Every chapter is full of useful tips, tricks, and pointers drawn from the real world. In addition to the various tables and lists in the text, we've collected these together in an appendix with pointers back into the book, so you can find things quickly when you need them.
In The CRC Card Book, we have tried to distill a wide range of experiences and training tips for the use of CRC. We hope that some of it applies to all of you, and that all of it applies to some of you. Above all, remember, let CRC make your team work fun!