Chapter 9 HANDLING EXCEPTIONS Outline v Learn about

Chapter 9 HANDLING EXCEPTIONS

Outline v Learn about the limitations of traditional error-handling methods v Throw exceptions v Use try blocks v Catch exceptions v Use multiple throw statements and multiple catch blocks v Use the default exception handler v Understand exception classes in the Standard Runtime Library v Derive your own exceptions from the exception class v Use exception specifications v Learn about unwinding the stack v Handle memory-allocation exceptions v Learn when to use exception handling 2

UNDERSTANDING THE LIMITATIONS OF TRADITIONAL ERROR HANDLING METHODS perform a first task if there was an error, perform some error processing else perform a second task if there was an error perform some error processing else perform a third task. . . and so on until the job is complete 3

UNDERSTANDING THE LIMITATIONS OF TRADITIONAL ERROR HANDLING METHODS v The error handling is interspersed with the logic of the program, making the program harder to read and maintain. § this example, each subsequent task is performed only if the previous task was successful, the number of nested decisions can become large, also making the program harder to read, and increasing the chance for mistakes. v The program is less efficient than it could be because decisions are constantly being made to test for situations that rarely occur. v Further complications make this approach unwieldy. The pseudocode contains several decisions “if there was an error” without indicating how the error is detected. v One possibility is that each of the tasks listed in the pseudocode is a function call, -> if each of the functions returns a code, then each function must be written so it returns nothing else 4

UNDERSTANDING THE LIMITATIONS OF TRADITIONAL ERROR HANDLING METHODS 5

UNDERSTANDING THE LIMITATIONS OF TRADITIONAL ERROR HANDLING METHODS v general rule of modular programming that provides the most flexibility is that a function should be able to determine an error situation, but not necessarily take action on it. v A better alternative to the function is to let the function discover the error, and then to notify the calling function of the error so the calling function can decide what to do about it. § -> flexibility by letting the calling function decide whether to carry on or terminate when the invalid data entry occurs. § This approach promotes function reusability—the get. User. Number() function can be used no matter what actions are needed after invalid data is entered. v Fortunately, object-oriented programming languages provide you with techniques that circumvent the problems of traditional error handling. -> exception handling; the actions you take with exceptions involve trying, throwing, and catching them. v try a function; if it throws an exception, you catch and handle it 6

THROWING EXCEPTIONS v A function can contain code to check for errors, and then send a message when it detects an error. v an exception is an object that contains information that is passed from the place where a problem occurs to another place that will handle the problem. § an exception object can be of any data type, including a scalar or class type, § and a variety of exception objects of different types can be sent from the same function, regardless of the function’s return type. § true to object-oriented style, you can create your own exception types that inherit from a built-in base exception class v any called function should check for errors, but should not be required to handle an error if one is found. v For example, suppose you call your hair stylist to make a 1 o’clock appointment on Thursday, and suppose the stylist has no openings then. v Maximum flexibility is achieved when the calling function is responsible for handling the error detected by the called function. 7

THROWING EXCEPTIONS v When an object-oriented program detects an error within a function, the function should send an error object to the calling function, or throw an exception. § A throw resembles a return statement in a function: does not continue from where the function was called. § It also resembles an exit()call, except that control does not have to be returned to the operating system. v Instead, when you throw an exception, control continues at a new program location chosen by the programmer where the program can catch, or receive, the exception that was thrown. 8

THROWING EXCEPTIONS 9

THROWING EXCEPTIONS 10

USING TRY BLOCKS v. A try block includes the following components: § the keyword try § an opening curly brace § the statements that are tried § a closing curly brace 11

USING TRY BLOCKS 12

CATCHING EXCEPTIONS v A catch block contains statements that execute when an exception is thrown and includes the following components: § the keyword catch § a single parameter in parentheses § an opening curly brace § one or more statements that describe the exception action to be taken § a closing curly brace 13

14

USING MULTIPLE THROW STATEMENTS AND MULTIPLE CATCH BLOCKS int main() { #include<iostream> int value; #include<string> try using namespace std; { int get. Positive. Single. Digit() value = get. Positive. Single. Digit(); { cout << "The entered value was good" << endl; int user. Entry; } catch(string msg) const int LOW = 0; { const int HIGH = 9; cout << "A message was thrown " << endl; const string ERROR_MSG = "Negative number"; cout << msg << endl; cout << "Enter a single-digit positive value "; } cin >> user. Entry; catch(int bad. Value) if(user. Entry < LOW) { const int REDUCTION = 10; throw(ERROR_MSG); value = bad. Value % REDUCTION; if(user. Entry >HIGH) cout << "A value was thrown " << endl; throw(user. Entry); cout << "The number you entered, " << bad. Value << ", is too return user. Entry; large. " << endl; } cout << "So it is being reduced to " << value << endl; } cout << "The value at the end of the program is “ << value << endl; return 0; } 15

DETERMINING THE ORDER OF catch BLOCKS v the first catch block that can accept a thrown object is the one that will execute. v When you create a function that throws several types, such as an integer and an Employee, it doesn’t matter which catch block you place first. v However, if you need to throw both a base class object and an object that is a member of its derived class from the same function: § For example, if you create an Employee class and a child Part. Time. Employee class, and either might be thrown from a function, then you must code the catch block for the Part. Time. Employee object first. 16

USING THE DEFAULT EXCEPTION HANDLER v When any object is thrown with a throw statement, then a subsequent catch block has a usable match if one of the following conditions is true: § The type of the thrown object and the type of the catch parameter are identical (for example, int and int). § The type of the thrown object and the type of the catch parameter are the same, int can be caught by const int, int&, or const int& § The catch parameter type is a parent class of the thrown argument. For example, if Part. Time. Employee derives from Employee, a catch block with an Employee parameter can catch both. v If you throw an exception and no catch block exists with an acceptable parameter type, then the program terminates -> code a default exception handler or default catch block that catches any type of object not previously caught. 17

UNDERSTANDING EXCEPTION CLASSES IN THE STANDARD RUNTIME LIBRARY 18

UNDERSTANDING EXCEPTION CLASSES IN THE STANDARD RUNTIME LIBRARY v logic_errors are errors in preconditions of some operation, such as passing an invalid value to a function v runtime_errors include problems that occur during program execution. v overflow_error occurs when the result of an arithmetic operation is larger than the largest number that can be stored on a computer 19

UNDERSTANDING EXCEPTION CLASSES IN THE STANDARD RUNTIME LIBRARY v A bad_alloc exception occurs when you attempt to allocate additional memory and an insufficient amount exists. v The dynamic_cast operator converts pointers from parent to child types. A bad_cast exception occurs when a dynamic_cast to a reference type fails. v A bad_exception is generated if you rethrow an exception from within an unexpected handler. You will learn about rethrowing exceptions later in this chapter. v The typeid operator returns information about an object’s data type. A bad_typeid exception is generated if you attempt to apply the typeid operator to a null expression. 20

AN EXAMPLE OF AN AUTOMATICALLY THROWN logic_error 21

USING THE what()FUNCTION 22

EXPLICITLY THROWING A BUILT-IN exception 23

24

DERIVING YOUR OWN EXCEPTIONS FROM THE EXCEPTION CLASS v making your programs easier for object-oriented programmers to understand. v Exception-handling using the built-in classes as a basis provides a more uniform way to handle errors. This is beneficial when many programmers combine multiple modules to create a large project. v If all exceptions derive from exception or one of its subclasses, then when you want to handle several program exceptions in the same way, you include a catch block with an exception parameter. v When an exception is of type exception or one of its subclasses, you can pass a string to the constructor that can be used as an informative message returned from the what() function. v override the parent class’s what()function with your own version and programmers who use your exception class will understand its purpose. 25

26

DERIVING YOUR OWN EXCEPTIONS FROM THE EXCEPTION CLASS catch(Range. Exception e) { cout << e. what() << endl; bad. User. Value = e. get. User(); if(bad. User. Value < LOW) { difference = LOW - bad. User. Value; cout << "The entry was " << difference << " too low" << endl; } else { difference = bad. User. Value - HIGH; cout << "The entry was " << difference <<" too high" << endl; } score = 0; } cout << "Score is " << score << endl; return 0; } #include<iostream> #include<exception> using namespace std; int main() { const int LOW = 3; const int HIGH = 7; int score; int bad. User. Value; int difference; try { score = get. Value. In. Range(LOW, HIGH); } 27

28

OVERRIDING THE EXCEPTION CLASS what()FUNCTION v Other programmers will immediately understand the purpose of and know how to use what()with exception objects. v If a client program catches a group of exceptions, such as all exceptions, or all runtime_exceptions, then one catch block can handle all of them, but use what() with the generic parameter to display the message appropriate to the caught subtype. 29

OVERRIDING THE EXCEPTION CLASS what()FUNCTION 30

USING EXCEPTION SPECIFICATIONS v how many different types of objects a function throws if you fail to carefully examine the code v You can explicitly indicate the exceptions that a function can possibly throw by writing an exception specification, which is a declaration of a function’s possible throw types v An exception specification is also called a throw list § int data. Entry() throw(char, double, Employee) § int data. Entry() throw() 31

EXCEPTION SPECIFICATIONS IN ANSI C++ v. American National Standards Institute (ANSI) § In ANSI C++, if a function throws an error whose type was not listed in its exception specification, then it will produce a run-time error called an unexpected exception, and abort the program. 32

HOW VISUAL C++ DEPARTS FROM THE ANSI STANDARD vthrow()means the function does not throw an exception vthrow(. . . )means the function can throw an exception vthrow(type)means the function can throw an exception of type 33

UNWINDING THE STACK v if function A calls function B, function B calls function C, and function C throws an exception, then if B does not catch the exception, function A can. If no HANDLING EXCEPTIONS function catches the exception, then the program terminates. This process is called unwinding the stack 34

UNWINDING THE STACK 35

36

37

38

39

RETHROWING EXCEPTIONS 40

HANDLING MEMORY ALLOCATION EXCEPTIONS 41

WHEN TO USE EXCEPTION HANDLING 42