CSE687 – Object Oriented Design Class Notes Chapter 1 - Modules Jim Fawcett copyright (c) 1994 - 2002 Chapter 1 - Modules

Definition Module: A module Chapter 1 - Modules

Limitation of Single File Program What’s wrong with a single file program? nothing, but as programs grow larger, this format becomes very limiting all functions are at the same (global) level; that’s what C does, you don’t have a choice programs grow too large to comprehend as a single entity eventually, as file size grows, the compiler can not swallow the whole file Chapter 1 - Modules

Limitation of Single File Program Solution: break programs up into modules In C++ each module consists of two files: a header file contains all public declarations so other modules know how to use this module’s services an implementation file contains all data and function definitions; that is, it provides the services that the header file announces Chapter 1 - Modules

Modularity Client Chapter 1 - Modules “In object oriented languages classes and objects form the logical structure of the system; we place these abstractions in modules to produce the system's physical architecture.” Grady Booch, Object Oriented Design with Applications, Benjamin/Cimmings, 1991 Modularization consists of dividing a program into modules which can be compiled separately. C++ performs type checking across module boundaries, using included header file declarations. Modules in C and C++ are simply separately compiled files. We place module interface declarations in header files. Module implementations are placed in separate files which include the header file at compilation time via a preprocessor #include “mod_name.hpp” directive. Public Interface header file mod_name.hpp Client Implementation client_name.cpp Module Implementation mod_name.cpp #include “mod_name.hpp” ... #include “mod_name.hpp” void main() { ... } #ifdef TEST_mod_name void main() { ... } #endif client_name module mod_name module Chapter 1 - Modules

Information Cluster Model Type: abstract system model Logical View: An information cluster encapsulates complex, sensitive, global, or device dependent data, along with functions which manage the data, in a container with internals not accessible to client view. Public access is provided by a series of accessor and mutator functions. Clients don’t have access to private functions or data. Implementation: C module using file scope for encapsulation. All private functions and global data are qualified as static. C++ class using public, protected, and private keywords to implement public and protected interfaces. Each class (or small, intimately related group of classes) should be given its own module. Each module implementing an information cluster contains a manual page and maintenance page which describe the logical model for the cluster and its chronological modification history. public functions private functions private data Chapter 1 - Modules

Module Implementation With two C files: Implementation file (with extension .c) contains all definitions of functions and data. Global data and all private functions are qualified with the static keyword. Every server module has a test stub in the implementation file (usually embodied in a main function) which is conditionally compiled when testing the module in isolation. Header file (with extension .h) which contains the declarations of all public functions, data structures, and macros. No data elements should be public. or with two C++ files: Implementation file (with extension .cpp) which contains definitions of member functions and static member data. Server modules have test stubs incorporated in their implementation files. Application modules contain a main function for the application and do not need a test stub. Header file (with extension .h) which contains class declaration(s). Public Interface functions go in the public sections of the declaration. Protected and private functions go in the corresponding sections. All data is defined as private member data in the private section of the class declaration. Chapter 1 - Modules

Module Implementation Each module begins with preprocessor guard statements which prevent multiple declarations, e.g.: #ifndef MODNAME_HPP #define MODNAME_HPP : header body : #endif The endif statement is the last in the header file. The test stub is the last part of implementation file and is also guarded to permit compilation without the stub: #ifdef TEST_MODNAME int main(int argc, char* argv[]) { : return err_code; } #endif Each module’s header file begins with a manual page and maintenance page: Manual page describes the logical view of module and provides brief descriptions of all public services provided by the module Maintenance page provides a chronological record of changes to the module since its creation, cited by version number. Either header file or implementation file may also contain a design notes page which describe the organizing principles, major data structures, and key processes which set the course of the module’s design. Chapter 1 - Modules

Module Structure Header File: modname.hpp #ifndef MODNAME_HPP #define MODNAME_HPP - manual page goes here - maintenance page goes here - declarations go here #endif Implementation File: modname.cpp - prologue goes here #include “modname.hpp” - function definitions go here #ifdef TEST_MODNAME void main() { - test code goes here } Chapter 1 - Modules

What goes in a Module’s Files? Elements of header file: Manual page Maintenance page Declarations of the module’s public classes and public global functions Definitions of inline functions, e.g., inline function bodies Function bodies of public template-based member functions and global functions Definitions for public constant data items Includes of other header files with declarations of types used in this header – no other files should be included. Elements of Implementation File Prologue, e.g., top part of the manual page, listing file name, platform, and author. Includes of header files declaring types needed by the code in this implementation – no other files should be included. Test stub – a main function which tests each function implemented in this file. It is also used to illustrate how a client would use the module and what its outputs look like. The test stub must be enclosed by compilation guards which enable compilation only when TEST_MODNAME is defined. Chapter 1 - Modules

Incremental Development Model Begins with requirements analysis, development of architecture and preliminary design. Result should be a design concept and partitioning into modules. As soon as modules are defined, one is selected that has no dependencies on other modules (at least one almost always exists). This module’s development proceeds by implementing one or two functions, adding tests of the new capabilities to the test stub, and verifying nominal operation. This process continues iteratively until module is complete. Detailed unit test then begins. Its goals are to exercise all paths and predicates in code to find all latent errors, correct them, and verify. Next a module is selected which depends at most on the tested module. It is subjected to same process to develop a complete and verified module. It is then integrated with the first module. This continues until all the modules have been integrated. Development completes by carrying out qualification test to demonstrate that software meets all its contractual obligations. The outstanding virtue of this approach is that only a small amount of new code is brought into a throughly tested baseline at each stage of development. Chapter 1 - Modules

Incremental Development Req. Anal. Preliminary Design Design Design --- Design phases Code & UT Code & UT --- Code & UT partially tested code Integration Test partially tested code correct code Integration Test correct code Integration Test modules Qualification Test Chapter 1 - Modules

Modular System Model Type: abstract system model Logical Model: collection of information clusters communicating through public interfaces. Implementation: One application module and a series of server modules. One C or C++ server module for each major program activity. The application module is the executive coordinator. Each server module has: an implementation file containing a test stub, with conditional compilation controls for stand alone testing. a header file which announces the module’s services to clients, e.g., the other servers and/or the application module. application.cpp server#1.hpp server#2.hpp server#1.cpp test stub #1 server#2.cpp test stub #2 Chapter 1 - Modules

CATALOG Modules Chapter 1 - Modules

Limitations of Modules What’s wrong with modular decomposition? nothing, but this format is not very flexible modules are defined statically (at compile time) new instances can not be created at run time this means that the flexability afforded by run-time creation of data does not extend to modules which contain both functions and data Solution: classes support declaration of objects which can be defined either statically or dynamically classes define both functions and data. An object, which is simply an instance of a class, can be defined statically in static memory, in local memory (on the stack), or dynamically on the heap, just like any intrinsic data type a program can define as many objects as it needs up to the amount that your computer memory can hold The classes you define will, of course, be packaged in modules. Chapter 1 - Modules

Module Example In the following pages we present the source code for fileInfo, one of the CATALOG program’s modules. Chapter 1 - Modules

CATALOG Classes Chapter 1 - Modules