notes6

A "do" loop has just one difference from a "while" loop. If Java reaches a "do" loop, it always executes the loop body at least once. Java doesn't check the loop condition until the end of the first iteration. "do" loops are appropriate for any loop you always want executed at least once, especially if the variables in the condition won't have meaningful assignments until the loop body has been executed.

do {
  s = keybd.readLine();
  process(s);
} while (s.length() > 0);               // Exit loop if s is an empty String.

A "break" statement immediately exits the innermost loop or "switch" statement enclosing the "break", and continues execution at the code following the loop or "switch".

In the loop example above, we might want to skip "process(s)" when s is a signal to exit (in this case, an empty String). We want a "time-and-a-half" loop–we want to enter the loop at a different point in the read-process cycle than we want to exit the loop at. Here are two alternative loops that do the right thing. They behave identically. Each has a different disadvantage.

  s = keybd.readLine();                  |  while (true) {       // Loop forever.
  while (s.length() > 0) {               |    s = keybd.readLine();
    process(s);                          |    if (s.length() == 0) {      
    s = keybd.readLine();                |      break;
  }                                      |    }                           
                                         |    process(s);
                                         |}
/*Disadvantage:  The line "s = keybd..." |
is repeated twice.  It's not really      |
a disadvantage here, but if input        | Disadvantage:  Somewhat obfuscated for 
took 100 lines of code, the              | the reader, because the loop isn't 
duplication would make the code harder   | aligned with its natural endpoint. 
to maintain.  Why?  Because a 
programmer improving the code might change one copy of the duplicated code 
without noticing the need to change the other to match.*/

Some loops have more than one natural endpoint. Suppose we want to iterate the read-process loop at most ten times. In the example at left below, the "break" statement cannot be criticized, because the loop has two natural endpoints. We could get rid of the "break" by writing the loop as at right below, but the result is longer and harder to read.

for (int i = 0; i < 10; i++) {       |  int i = 0;
  s = keybd.readLine();              |  do {
  if (s.length() == 0) {             |    s = keybd.readLine();       
    break;                           |    if (s.length() > 0) {             
  }                                  |      process(s);                      
  process(s);                        |    }
}                                    |    i++;
                                     |  } while ((i < 10) &&
                                     |           (s.length() > 0));

There are anti-break zealots who claim that the loop on the right is the "correct" way to do things. I disagree, because the left loop is clearly more readable.

Some of the zealots feel this way because "break" statements are a little bit like the "go to" statements found in some languages like Basic and Fortran (and the machine language that microprocessors really execute). "go to" statements allow you to jump to any line of code in the program. It sounds like a good idea at first, but it invariably leads to insanely unmaintainable code. For example, what happens if you jump to the middle of a loop? Turing Award winner Edsger Dijkstra wrote a famous article in 1968 entitled "Go To Statement Considered Harmful", which is part of the reason why many modern languages like Java don't have "go to" statements.

Both "break" and "return" are limited forms of "go to" statements. Their limitations prohibit the worst abuses of "go to". They allow control flow to jump in your program in ways that are straightforward to understand.

WARNING: It's easy to forget exactly where a "break" statement will jump to. For example, "break" does not jump to the end of the innermost enclosing "if" statement. An AT&T programmer introduced a bug into telephone switching software in a procedure that contained a "switch" statement, which contained an "if" clause, which contained a "break", which was intended for the "if" clause, but instead jumped to the end of the "switch" statement. As a result, on January 15, 1990, AT&T's entire U.S. long distance service collapsed for eleven hours. (That code was actually written in C, but Java and C use identical syntax and semantics for loops, "switch", and "break".)

The "continue" statement is akin to the "break" statement, except (1) it only applies to loops, and (2) it jumps to the end of the loop body but it doesn't necessarily exit the loop; another iteration will commence if the loop condition is satisfied.

Finally, I told you that "for" loops are identical to certain "while" loops, but there's actually a subtle difference when you use "continue". What's the difference between the following two loops?

int i = 0;                           | for (int i = 0; i < 10; i++) {
while (i < 10) {                     |   if (condition(i)) {
  if (condition(i)) {                |     continue;
    continue;                        |   }
  }                                  |   call(i);
  call(i);                           | }
  i++;                               |
}                                    |

Answer: when "continue" is called in the "while" loop, "i++" is not executed. In the "for" loop, however, i is incremented at the end of every iteration, even iterations where "continue" is called.