Class frameworks are the software commodities of the future—Learn a methodology today to develop them for tomorrow!
The growth of standards in the software industry is enabling designers to more effectively use the Object Paradigm to create reusable solutions for common business problems. These designers, however, must use a radically different development methodology—the methodology you'll learn from this book.
First you'll come to understand what class frameworks are and the software standards—STL, CORBA, and ODMG-93—on which they are built. Then you'll discover how to develop requirements for a framework by performing domain analysis.
Next you'll learn how to develop a framework with:
Two case studies illustrate the methodology being applied to the development of a horizontal and a vertical framework. When you've mastered the concepts in Framework-Based Software Development in C++, you'll be prepared to develop a repository of reusable software that is based on open standards and off-the-shelf products. This will make your software organization far more competitive. If you are a software entrepreneur, it could also make you far more profitable.
I. INTRODUCTION.1. Overview of Methodology.
What Is a Framework? What Are the Benefits of Using Frameworks? What Is the Difference Between a Horizontal and Vertical Framework? What Is Framework-Based Software Development? What Is the Object Infrastructure?2. The Standard Template Library (STL).
Getting Started with STL. The Design of STL. Using the string Class. Using STL Container Classes. Adding Your Own Generic Algorithms.3. Common Object Request Broker Architecture (CORBA).
High-Level Overview. The Marriage Between the WWW and CORBA. Using an ORB-An Informal Description. Using an ORB-The Detailed Steps. CORBA and Frameworks. The Impact of CORBA on Modeling Notation. An Example: Event Notification. Common Object Services.4. ODMG-93: An Object Database Standard.
High-Level Overview. The Role of Object Databases in Frameworks. The Impact of Object Databases on Modeling Notation. Using an ODMG-Compliant Object Database with C++. Example Application Using CORBA and ODMG-93: Security Alarm Reporting.
II. METHODOLOGY.5. Domain Analysis.
Steps in Domain Analysis. An Example Domain Analysis for “Mail Order”.6. Framework Design Standard.
Introduction to Design Patterns. Catalog of Design Patterns. Design Rules.7. Framework Design Metrics.
Inheritance Depth Metric. Inheritance Width Metric. Size Metrics. Framework Reusefulness Metric.8. Framework Development Procedures.
Designing a Framework. Implementing a Framework. Testing a Framework. Preparing User Documentation.9. Framework Development Strategies.
Creating a Framework Industry. Reengineering the Development Organization. Putting Together an Application Development Team. Putting Together an Infrastructure Group. Evaluating an Off-the-Shelf Framework. The Economics of In-house Framework Development. Setting Up the Ideal Development Environment. The Project Plan for Developing a Framework. Red Flags.
III. EXAMPLE FRAMEWORKS.10. A Horizontal Framework for Workflow.
Facet Analysis: Workflow. Design of the Workflow Framework. Implementation of the Workflow Framework.11. A Framework for Monitoring Financial Risk.
Domain Analysis: Financial Risk Monitoring. Design of the Risk Monitoring Framework.APPENDIX A: STL Lint Program.
Rogers, FRAMEWORK-BASED SOFTWARE DEVELOPMENT IN C++ 1/E ISBN 0-13-533365-2 PREFACE
This is a handbook for building the next generation of software commodities, called frameworks, on an object infrastructure consisting of CORBA¿, ODMG-93 (a standard object database API), and ANSI C++ with STL. The development methodology focuses on domain analysis and a set of design patterns that are well suited to the development of frameworks that must operate in a distributed environment. Before I talk more about the book, I think it is important to give you a little history about the circumstances that compelled me to write it.
I've been a software developer for about fifteen years now, and it's just no fun lately. I'm so tired of having to rebuild proprietary software infrastructure for each new project. I'd rather be spending my time solving more complex problems. Luckily, there are credible standards bodies that have put a tremendous amount of effort into defining a software infrastructure based on the Object paradigm. As this infrastructure matures, it is becoming possible for skilled developers to finally take on more complex problems. Some of us will be able to do so in large development organizations. Many of us, however, will have the opportunity to become entrepreneurs.
I wasn't always optimistic about my future in software development. A few years ago I was almost embarrassed to say that I was an object-oriented designer. It all started when the OO companies began to emerge in the early 1990s. With them came the armies of sales and marketing geniuses promising all kinds of great things about OO technology and C++. Companies everywhere invested in OO/C++ tools, OO/C++ consultants, and OO/C++ training, so that they wouldn't miss out on achieving the tremendous levels of software reuse promised. This would help them "gain the competitive edge," "increase profit margins," and stuff like that. Well, by the mid 1990s none of this had really materialized and corporate America was growing increasingly impatient. It was just about then that I started to hear people condemning C++ for its complexity.
The root cause of all the disappointment was the widely held belief that OO would allow people of all skill levels to approach hard problems from an easier angle. It doesn't. On the contrary, using the paradigm to its full advantage demands that a hard problem be approached from an even harder angle. Many people didn't understand this. Those who did were often not given the time to use the paradigm correctly.
All this nonsense didn't discourage people in standards organizations like ANSI, the OMG®, and the ODMG that used the period of OO-mania to create standards that are so essential to the case for C++ and OO. As a result of their efforts, there is now a tidal wave of object standards approaching. Because of the standards efforts, developers can now buy an industry-standard "object infrastructure" for less than the cost of a good PC, and it's only going to get cheaper. This is really exciting because it makes possible, and even encourages, a new OO-related industry to emerge. This new industry would thrive on a mixture of highly skilled software entrepreneurs that specialize in solving complex problems using OO, and corporate customers who have complex problems to solve, but have not had the resources to solve them. The entrepreneurs would focus on building software that captures reusable abstractions like business models, analytical models, industrial processes, and scientific phenomenon. These software "frameworks" could become commodities designed to work on the object infrastructure.
The challenging development work in the years ahead will be done, in large part, by two types of software developers: those who develop applications using frameworks and those who actually develop the frameworks. The framework users will be average programmers with an in-depth knowledge of specific problems. The framework developers will be extraordinary programmers, with tremendous breadth of domain knowledge and an uncanny ability to generalize a situation. You need to start to decide what you want your role to be in this new order. If you have ambitions to be a framework developer, you need to start to cultivate the appropriate skills. And now I arrive at the reason for writing this book, and the reason you may want to read it. To my knowledge, there is no publicly documented methodology that explains how to develop software frameworks in C++. Reading this book is a good first step toward becoming a framework developer.
Overview of Contents
The book has three parts: Part 1 - Introduction, Part 2 - Methodology, and Part 3 - Example Frameworks. Part 1 starts by giving an overview of the methodology, which I call "Framework-Based Software Development." In order to understand the details of the methodology, it is necessary to have a basic knowledge of the new standards comprising the object infrastructure that the methodology depends on: ANSI C++ with STL, CORBA, and ODMG-93. I, therefore, include a chapter for each of the new standards. I should mention that I am not going to explain any features of the core C++ language; only STL, because it is so new. If you aren't up on some of the newer language features, like namespace, for instance, you may have to do some homework. But, wait until you encounter the unfamiliar features before you run out and buy another book. You may be able to understand the feature just by seeing the way I use it.
The infrastructure offers tremendous productivity increases over the proprietary ones we've had to use in the past. As you will see, however, using such an infrastructure to build frameworks requires skills that are more sophisticated than has traditionally been required. Framework developers will also be data modelers, DBAs, analysts, and system architects.
Part 2 of the book presents the actual methodology. The methodology includes:
7Procedures for creating framework requirements
7The framework design standard
7Framework design metrics
7Procedures that show you how to design and implement a conforming framework
7Recommendations for documenting and testing frameworks
7Recommended strategies for managing a software development organization
In Part 3 of the book, I use the methodology to develop two frameworks. I will walk through the analysis, design, and some of the implementation of each of these frameworks and illustrate how they would be used to develop applications.
I am grateful to Paul Becker, my publisher, for making this project possible. I would like to thank Frank Gielen, Chak Kolli, John Peterson, Michael Connick, Sal Ricci, David Spuler, John McEnroe, and P. J. Plauger for their comments on the manuscript. Paul Hickey and John Petri from IONA Technologies have been very supportive. Mike Florence and Sophie Gamerman from O2 Technologies were also very helpful.
This book exists because my good friend and former colleague Joe DiBiase convinced me to write it and encouraged me all the way through it. Thanks, Joe.
I need to apologize to my kids-Megan, Lauren, Kierstin, and Glenn-for all the time I missed with them over the past year. And finally, I must thank my wife, Kim, for doing her best to keep the kids occupied as I wrote the manuscript in the corner of my living room.
Greg Rogers Point Pleasant, New Jersey firstname.lastname@example.org