“And if you take one from three hundred and sixty-five, what remains?”
“Three hundred and sixty-four, of course.”
Humpty Dumpty looked doubtful. “I’d rather see that done on paper,” he said.
LEWIS CARROLL, Through the Looking-Glass
A mistake in a program is usually called a bug, and the process of eliminating bugs is called debugging. There is colorful history of how this term came into use. It occurred in the early days of computers, when computer hardware was extremely sensitive and occupied an entire room. Rear Admiral Grace Murray Hopper (1906–1992) was “the third programmer on the world’s first large-scale digital computer.” (Denise W. Gurer, “Pioneering women in computer science” CACM 38(1):45–54, January 1995.) While Hopper was working on the Harvard Mark I computer under the command of Harvard professor Howard H. Aiken, an unfortunate moth caused a relay to fail. Hopper and the other programmers taped the deceased moth in the logbook with the note “First actual case of bug being found.” The logbook is currently on display at the Naval Museum in Dahlgren, Virginia. This was the first documented computer bug. Professor Aiken would come into the facility during a slack time and inquire if any numbers were being computed. The programmers would reply that they were debugging the computer. For more information about Admiral Hopper and other persons in computing, see Robert Slater, Portraits in Silicon (MIT Press, 1987). Today, a bug is a mistake in a program. In this section we describe the three main kinds of programming mistakes and give some hints on how to correct them.
The compiler will catch certain kinds of mistakes and will write out an error message when it finds a mistake. It will detect what are called syntax errors, because they are, by and large, violation of the syntax (that is, the grammar rules) of the programming language, such as omitting a semicolon.
If the compiler discovers that your program contains a syntax error, it will tell you where the error is likely to be and what kind of error it is likely to be. When the compiler says your program contains a syntax error, you can be confident that it does. However, the compiler may be incorrect about either the location or the nature of the error. It does a better job of determining the location of an error, to within a line or two, than it does of determining the source of the error. This is because the compiler is guessing at what you meant to write down and can easily guess wrong. After all, the compiler cannot read your mind. Error messages subsequent to the first one have a higher likelihood of being incorrect with respect to either the location or the nature of the error. Again, this is because the compiler must guess your meaning. If the compiler’s first guess was incorrect, this will affect its analysis of future mistakes, since the analysis will be based on a false assumption.
If your program contains something that is a direct violation of the syntax rules for your programming language, the compiler will give you an error message. However, sometimes the compiler will give you only a warning message, which indicates that you have done something that is not, technically speaking, a violation of the programming language syntax rules, but that is unusual enough to indicate a likely mistake. When you get a warning message, the compiler is saying, “Are you sure you mean this?” At this stage of your development, you should treat every warning as if it were an error until your instructor approves ignoring the warning.
There are certain kinds of errors that the computer system can detect only when a program is run. Appropriately enough, these are called run-time errors. Most computer systems will detect certain run-time errors and output an appropriate error message. Many run-time errors have to do with numeric calculations. For example, if the computer attempts to divide a number by zero, that is normally a run-time error.
If the compiler approved of your program and the program ran once with no run-time error messages, this does not guarantee that your program is correct. Remember, the compiler will only tell you if you wrote a syntactically (that is, grammatically) correct C++ program. It will not tell you whether the program does what you want it to do. Mistakes in the underlying algorithm or in translating the algorithm into the C++ language are called logic errors. For example, if you were to mistakenly use the addition sign + instead of the multiplication sign * in the program in Display 1.8, that would be a logic error. The program would compile and run normally but would give the wrong answer. If the compiler approves of your program and there are no run-time errors but the program does not perform properly, then undoubtedly your program contains a logic error. Logic errors are the hardest kind to diagnose, because the computer gives you no error messages to help find the error. It cannot reasonably be expected to give any error messages. For all the computer knows, you may have meant what you wrote.
What are the three main kinds of program errors?
What kinds of errors are discovered by the compiler?
If you omit a punctuation symbol (such as a semicolon) from a program, an error is produced. What kind of error?
Omitting the final brace } from a program produces an error. What kind of error?
Suppose your program has a situation about which the compiler reports a warning. What should you do about it? Give the text’s answer and your local answer if it is different from the text’s. Identify your answers as the text’s or as based on your local rules.
Suppose you write a program that is supposed to compute the interest on a bank account at a bank that computes interest on a daily basis, and suppose you incorrectly write your program so that it computes interest on an annual basis. What kind of program error is this?