A thread is a program unit that is executed independently of other parts of the program. The Java virtual machine executes each thread for a short amount of time and then switches to another thread. This gives the illusion of executing the threads in parallel to each other. Actually, if a computer has multiple central processing units (CPUs), then some of the threads can run in parallel, one on each processor.

Running a thread is simple in Java—follow these steps:

Let us look at a concrete example. We want to print ten greetings of "Hello, World!", one greeting every second. We will add a time stamp to each greeting to see when it is printed.

Mon Dec 28 23:12:03 PST 2009 Hello, World!
Mon Dec 28 23:12:04 PST 2009 Hello, World!
Mon Dec 28 23:12:05 PST 2009 Hello, World!
Mon Dec 28 23:12:06 PST 2009 Hello, World!
Mon Dec 28 23:12:07 PST 2009 Hello, World!
Mon Dec 28 23:12:08 PST 2009 Hello, World!
Mon Dec 28 23:12:09 PST 2009 Hello, World!
Mon Dec 28 23:12:10 PST 2009 Hello, World!
Mon Dec 28 23:12:11 PST 2009 Hello, World!
Mon Dec 28 23:12:12 PST 2009 Hello, World!

Using the instructions for creating a thread, define a class that implements the Runnable interface:

public class GreetingRunnable implements Runnable
{
   private String greeting;

   public GreetingRunnable(String aGreeting)
   {
       greeting = aGreeting;
   }

   public void run()
   {
      Task statements
      ...
   }
}

The run method should loop ten times through the following task actions:

Get the time stamp by constructing an object of the java.util.Date class. Its default constructor produces a date that is set to the current date and time.

Date now = new Date();
System.out.println(now + " " + greeting);

To wait a second, we use the static sleep method of the Thread class. The call

Thread.sleep(milliseconds)

puts the current thread to sleep for a given number of milliseconds. In our case, it should sleep for 1,000 milliseconds, or one second.

There is, however, one technical problem. Putting a thread to sleep is potentially risky—a thread might sleep for so long that it is no longer useful and should be terminated. As you will see in Section 20.2, to terminate a thread, you interrupt it. When a sleeping thread is interrupted, an InterruptedException is generated. You need to catch that exception in your run method and terminate the thread.

The simplest way to handle thread interruptions is to give your run method the following form:

public void run()
{
   try
   {
      Task statements

}
   catch (InterruptedException exception)
   {
   }
   Clean up, if necessary
}

We follow that structure in our example. Here is the complete code for our runnable class:

ch20/greeting/GreetingRunnable.java

1  import java.util.Date;
 2
 3  /**
 4     A runnable that repeatedly prints a greeting.
 5  */
 6  public class GreetingRunnable implements Runnable
 7  {
 8     private static final int REPETITIONS = 10;
 9     private static final int DELAY = 1000;
10
11     private String greeting;
12
13     /**
14        Constructs the runnable object.
15        @param aGreeting the greeting to display
16     */
17     public GreetingRunnable(String aGreeting)
18     {
19        greeting = aGreeting;
20     }
21
22     public void run()
23     {
24        try
25        {
26           for (int i = 1; i <= REPETITIONS; i++)
27           {
28              Date now = new Date();
29              System.out.println(now + " " + greeting);
30              Thread.sleep(DELAY);
31           }
32        }
33        catch (InterruptedException exception)
34        {
35        }
36     }
37  }

To start a thread, first construct an object of the runnable class.

Runnable r = new GreetingRunnable("Hello, World!");

Then construct a thread and call the start method.

Thread t = new Thread(r);
t.start();

Now a new thread is started, executing the code in the run method of your runnable class in parallel with any other threads in your program.

In the GreetingThreadRunner program, we start two threads: one that prints "Hello, World!" and one that prints "Goodbye, World!"

ch20/greeting/GreetingThreadRunner.java

1  /**
 2     This program runs two greeting threads in parallel.
 3  */
 4  public class GreetingThreadRunner
 5  {
 6     public static void main(String[] args)
 7     {
 8        GreetingRunnable r1 = new GreetingRunnable("Hello, World!");
 9        GreetingRunnable r2 = new GreetingRunnable("Goodbye, World!");
10        Thread t1 = new Thread(r1);
11        Thread t2 = new Thread(r2);
12        t1.start();
13        t2.start();
14     }
15  }

Program Run

Mon Dec 28 12:04:46 PST 2009 Hello, World!
Mon Dec 28 12:04:46 PST 2009 Goodbye, World!
Mon Dec 28 12:04:47 PST 2009 Hello, World!
Mon Dec 28 12:04:47 PST 2009 Goodbye, World!
Mon Dec 28 12:04:48 PST 2009 Hello, World!
Mon Dec 28 12:04:48 PST 2009 Goodbye, World!
Mon Dec 28 12:04:49 PST 2009 Hello, World!
Mon Dec 28 12:04:49 PST 2009 Goodbye, World!
Mon Dec 28 12:04:50 PST 2009 Hello, World!
Mon Dec 28 12:04:50 PST 2009 Goodbye, World!
Mon Dec 28 12:04:51 PST 2009 Hello, World!
Mon Dec 28 12:04:51 PST 2009 Goodbye, World!
Mon Dec 28 12:04:52 PST 2009 Goodbye, World!
Mon Dec 28 12:04:52 PST 2009 Hello, World!
Mon Dec 28 12:04:53 PST 2009 Hello, World!
Mon Dec 28 12:04:53 PST 2009 Goodbye, World!
Mon Dec 28 12:04:54 PST 2009 Hello, World!
Mon Dec 28 12:04:54 PST 2009 Goodbye, World!
Mon Dec 28 12:04:55 PST 2009 Hello, World!
Mon Dec 28 12:04:55 PST 2009 Goodbye, World!

Because both threads are running in parallel, the two message sets are interleaved. However, if you look closely, you will find that the two threads aren't exactly inter-leaved. Sometimes, the second thread seems to jump ahead of the first thread. This shows an important characteristic of threads. The thread scheduler gives no guarantee about the order in which threads are executed. Each thread runs for a short amount of time, called a time slice. Then the scheduler activates another thread. However, there will always be slight variations in running times, especially when calling operating system services (such as input and output). Thus, you should expect that the order in which each thread gains control is somewhat random.

public class MyThread extends Thread
{
   public void run()
   {
      Task statements
      ...
   }
}

Thread t = new MyThread();
t.start();

Runnable r1 = new GreetingRunnable("Hello, World!");
Runnable r2 = new GreetingRunnable("Goodbye, World!");
ExecutorService pool = Executors.newFixedThreadPool(MAX_THREADS);

pool.execute(r1);
pool.execute(r2);

A thread terminates when the run method of the associated runnable object returns. This is the normal way of terminating a thread—implement the run method so that it returns when it determines that no more work needs to be done.

However, sometimes you need to terminate a running thread. For example, you may have several threads trying to find a solution to a problem. As soon as the first one has succeeded, you may want to terminate the other ones. In the initial release of the Java library, the Thread class had a stop method to terminate a thread. However, that method is now deprecated—computer scientists have found that stopping a thread can lead to dangerous situations when multiple threads share objects. (We will discuss access to shared objects in Section 20.3.) Instead of simply stopping a thread, you should notify the thread that it should be terminated. The thread needs to cooperate, by releasing any resources that it is currently using and doing any other required cleanup. In other words, a thread should be in charge of terminating itself.

To notify a thread that it should clean up and terminate, you use the interrupt method.

t.interrupt();

This method does not actually cause the thread to terminate—it merely sets a boolean variable in the thread data structure.

The run method can check whether that flag has been set, by calling the static interrupted method. In that case, it should do any necessary cleanup and exit. For example, the run method of the GreetingRunnable could check for interruptions at the beginning of each loop iteration:

public void run()
{
   for (int i = 1;
        i <= REPETITIONS && !Thread.interrupted();
        i++)
   {
      Do work
   }
   Clean up
}

However, if a thread is sleeping, it can't execute code that checks for interruptions. Therefore, the sleep method is terminated with an InterruptedException whenever a sleeping thread is interrupted. The sleep method also throws an InterruptedException when it is called in a thread that is already interrupted. If your run method calls sleep in each loop iteration, simply use the InterruptedException to find out whether the thread is terminated. The easiest way to do that is to surround the entire work portion of the run method with a try block, like this:

public void run()
{
   try
   {
      for (int i = 1; i <= REPETITIONS; i++)
      {
         Do work
         Sleep
      }
   }
   catch (InterruptedException exception)
   {
   }
   Clean up
}

Strictly speaking, there is nothing in the Java language specification that says that a thread must terminate when it is interrupted. It is entirely up to the thread what it does when it is interrupted. Interrupting is a general mechanism for getting the thread's attention, even when it is sleeping. However, in this chapter, we will always terminate a thread that is being interrupted.

public void run()
{
   while (...)
   {
       ...
      try
      {
         Thread.sleep(delay);
      }
      catch (InterruptedException exception) {} // DON'T
       ...
   }
}

public void run()
{
   try
   {
      while (...)
      {
         ...
         Thread.sleep(delay);
         ...
      }
   }
   catch (InterruptedException exception) {} // OK
}

When threads share access to a common object, they can conflict with each other. To demonstrate the problems that can arise, we will investigate a sample program in which multiple threads manipulate a bank account.

We construct a bank account that starts out with a zero balance. We create two sets of threads:

Here is the run method of the DepositRunnable class:

public void run()
{
   try
   {
      for (int i = 1; i <= count; i++)
      {
         account.deposit(amount);
         Thread.sleep(DELAY);
      }
   }
   catch (InterruptedException exception)
   {
   }
}

The WithdrawRunnable class is similar—it withdraws money instead.

The deposit and withdraw methods of the BankAccount class have been modified to print messages that show what is happening. For example, here is the code for the deposit method:

public void deposit(double amount)
{
   System.out.print("Depositing " + amount);
   double newBalance = balance + amount;
   System.out.println(", new balance is " + newBalance);
   balance = newBalance;
}

You can find the complete source code at the end of this section.

Normally, the program output looks somewhat like this:

Depositing 100.0, new balance is 100.0
Withdrawing 100.0, new balance is 0.0
Depositing 100.0, new balance is 100.0
Depositing 100.0, new balance is 200.0
Withdrawing 100.0, new balance is 100.0
...
Withdrawing 100.0, new balance is 0.0

In the end, the balance should be zero. However, when you run this program repeatedly, you may sometimes notice messed-up output, like this:

Depositing 100.0Withdrawing 100.0, new balance is 100.0
, new balance is −100.0

And if you look at the last line of the output, you will notice that the final balance is not always zero. Clearly, something problematic is happening.

You may have to try the program several times to see this effect. Here is a scenario that explains how a problem can occur.

Figure 20.1. Corrupting the Contents of the balance Variable

Thus, not only are the messages interleaved, but the balance is wrong. The balance after a withdrawal and deposit should again be 0, not 100. Because the deposit method was interrupted, it used the old balance (before the withdrawal) to compute the value of its local newBalance variable. Later, when it was activated again, it used that newBalance value to overwrite the changed balance variable.

As you can see, each thread has its own local variables, but all threads share access to the balance instance variable. That shared access creates a problem. This problem is often called a race condition. All threads, in their race to complete their respective tasks, manipulate a shared variable, and the end result depends on which of them happens to win the race.

You might argue that the reason for this problem is that we made it too easy to interrupt the balance computation. Suppose the code for the deposit method is reorganized like this:

public void deposit(double amount)
{
   balance = balance + amount;
   System.out.print("Depositing " + amount
         + ", new balance is " + balance);
}

Suppose further that you make the same change in the withdraw method. If you run the resulting program, everything seems to be fine.

However, that is a dangerous illusion. The problem hasn't gone away; it has become much less frequent, and, therefore, more difficult to observe. It is still possible for the deposit method to reach the end of its time slice after it has computed the right-hand-side value

balance + amount

but before it performs the assignment

balance = the right-hand-side value

When the method regains control, it finally carries out the assignment, putting the wrong value into the balance variable.

ch20/unsynch/BankAccountThreadRunner.java

1  /**
 2     This program runs threads that deposit and withdraw
 3     money from the same bank account.
 4  */
 5  public class BankAccountThreadRunner
 6  {
 7     public static void main(String[] args)
 8     {
 9        BankAccount account = new BankAccount();
10        final double AMOUNT = 100;
11        final int REPETITIONS = 100;
12        final int THREADS = 100;
13
14        for (int i = 1; i <= THREADS; i++)
15        {
16           DepositRunnable d = new DepositRunnable(
17                  account, AMOUNT, REPETITIONS);
18           WithdrawRunnable w = new WithdrawRunnable(
19                  account, AMOUNT, REPETITIONS);
20

21           Thread dt = new Thread(d);
22           Thread wt = new Thread(w);
23
24           dt.start();
25           wt.start();
26        }
27     }
28  }

ch20/unsynch/DepositRunnable.java

1  /**
 2     A deposit runnable makes periodic deposits to a bank account.
 3  */
 4  public class DepositRunnable implements Runnable
 5  {
 6     private static final int DELAY = 1;
 7     private BankAccount account;
 8     private double amount;
 9     private int count;
10
11     /**
12        Constructs a deposit runnable.
13        @param anAccount the account into which to deposit money
14        @param anAmount the amount to deposit in each repetition
15        @param aCount the number of repetitions
16     */
17     public DepositRunnable(BankAccount anAccount, double anAmount,
18            int aCount)
19     {
20        account = anAccount;
21        amount = anAmount;
22        count = aCount;
23     }
24
25     public void run()
26     {
27        try
28        {
29           for (int i = 1; i <= count; i++)
30           {
31              account.deposit(amount);
32              Thread.sleep(DELAY);
33           }
34        }
35        catch (InterruptedException exception) {}
36     }
37  }

ch20/unsynch/WithdrawRunnable.java

1    /**
 2       A withdraw runnable makes periodic withdrawals from a bank account.
 3    */
 4    public class WithdrawRunnable implements Runnable
 5    {
 6       private static final int DELAY = 1;
 7       private BankAccount account;

8      private double amount;
 9      private int count;
10
11      /**
12          Constructs a withdraw runnable.
13          @param anAccount the account from which to withdraw money
14          @param anAmount the amount to withdraw in each repetition
15          @param aCount the number of repetitions
16      */
17      public WithdrawRunnable(BankAccount anAccount, double anAmount,
18              int aCount)
19      {
20          account = anAccount;
21          amount = anAmount;
22          count = aCount;
23      }
24
25      public void run()
26      {
27          try
28          {
29             for (int i = 1; i <= count; i++)
30             {
31                account.withdraw(amount);
32                Thread.sleep(DELAY);
33             }
34          }
35          catch (InterruptedException exception) {}
36      }
37   }

ch20/unsynch/BankAccount.java

1  /**
 2     A bank account has a balance that can be changed by
 3     deposits and withdrawals.
 4  */
 5  public class BankAccount
 6  {
 7     private double balance;
 8
 9     /**
10        Constructs a bank account with a zero balance.
11     */
12     public BankAccount()
13     {
14        balance = 0;
15     }
16
17     /**
18        Deposits money into the bank account.
19        @param amount the amount to deposit
20     */
21     public void deposit(double amount)
22     {
23        System.out.print("Depositing " + amount);
24        double newBalance = balance + amount;
25        System.out.println(", new balance is " + newBalance);
26        balance = newBalance;

27     }
28
29     /**
30        Withdraws money from the bank account.
31        @param amount the amount to withdraw
32     */
33     public void withdraw(double amount)
34     {
35        System.out.print("Withdrawing " + amount);
36        double newBalance = balance - amount;
37        System.out.println(", new balance is " + newBalance);
38        balance = newBalance;
39     }
40
41     /**
42          Gets the current balance of the bank account.
43          @return the current balance
44     */
45     public double getBalance()
46     {
47          return balance;
48     }
49  }

Program Run

Depositing 100.0, new balance is 100.0
Withdrawing 100.0, new balance is 0.0
Depositing 100.0, new balance is 100.0
Withdrawing 100.0, new balance is 0.0
...
Withdrawing 100.0, new balance is 400.0
Depositing 100.0, new balance is 500.0
Withdrawing 100.0, new balance is 400.0
Withdrawing 100.0, new balance is 300.0

6.Suppose two threads simultaneously insert objects into a linked list. Using the implementation in Chapter 15, explain how the list can be damaged in the process.

To solve problems such as the one that you observed in the preceding section, use a lock object. The lock object is used to control the threads that want to manipulate a shared resource.

The Java library defines a Lock interface and several classes that implement this interface. The ReentrantLock class is the most commonly used lock class, and the only one that we cover in this book. (Locks are a feature added in Java version 5.0. Earlier versions of Java have a lower-level facility for thread synchronization—see Special Topic 20.2 on page 823).

Typically, a lock object is added to a class whose methods access shared resources, like this:

public class BankAccount
{
   private Lock balanceChangeLock;
   ...
   public BankAccount()
   {
      balanceChangeLock = new ReentrantLock();
      ...
   }
}

All code that manipulates the shared resource is surrounded by calls to lock and unlock the lock object:

balanceChangeLock.lock();
Manipulate the shared resource
balanceChangeLock.unlock();

However, this sequence of statements has a potential flaw. If the code between the calls to lock and unlock throws an exception, the call to unlock never happens. This is a serious problem. After an exception, the current thread continues to hold the lock, and no other thread can acquire it. To overcome this problem, place the call to unlock into a finally clause:

balanceChangeLock.lock();
try
{
   Manipulate the shared resource
}
finally
{
   balanceChangeLock.unlock();
}

For example, here is the code for the deposit method:

public void deposit(double amount)
{
   balanceChangeLock.lock();
   try
   {
      System.out.print("Depositing " + amount);
      double newBalance = balance + amount;
      System.out.println(", new balance is " + newBalance);
      balance = newBalance;
   }
   finally
   {
      balanceChangeLock.unlock();
   }
}

When a thread calls the lock method, it owns the lock until it calls the unlock method. If a thread calls lock while another thread owns the lock, it is temporarily deactivated. The thread scheduler periodically reactivates such a thread so that it can again try to acquire the lock. If the lock is still unavailable, the thread is again deactivated. Eventually, when the lock is available because the original thread unlocked it, the waiting thread can acquire the lock.

Figure 20.2. Visualizing Object Locks

One way to visualize this behavior is to imagine that the lock object is the lock of an old-fashioned telephone booth and the threads are people wanting to make telephone calls (see Figure 2). The telephone booth can accommodate only one person at one time. If the booth is empty, then the first person wanting to make a call goes inside and closes the door. If another person wants to make a call and finds the booth occupied, then the second person needs to wait until the first person leaves the booth. If multiple people want to gain access to the telephone booth, they all wait outside. They don't necessarily form an orderly queue; a randomly chosen person may gain access when the telephone booth becomes available again.

With the ReentrantLock class, a thread can call the lock method on a lock object that it already owns. This can happen if one method calls another, and both start by locking the same object. The thread gives up ownership if the unlock method has been called as often as the lock method.

By surrounding the code in both the deposit and withdraw methods with lock and unlock calls, we ensure that our program will always run correctly. Only one thread at a time can execute either method on a given object. Whenever a thread acquires the lock, it is guaranteed to execute the method to completion before the other thread gets a chance to modify the balance of the same bank account object.

8.What happens if we omit the call unlock at the end of the deposit method?

You can use lock objects to ensure that shared data are in a consistent state when several threads access them. However, locks can lead to another problem. It can happen that one thread acquires a lock and then waits for another thread to do some essential work. If that other thread is currently waiting to acquire the same lock, then neither of the two threads can proceed. Such a situation is called a deadlock or deadly embrace. Let's look at an example.

Suppose we want to disallow negative bank balances in our program. Here's a naive way of doing that. In the run method of the WithdrawRunnable class, we can check the balance before withdrawing money:

if (account.getBalance() >= amount)
   account.withdraw(amount);

This works if there is only a single thread running that withdraws money. But suppose we have multiple threads that withdraw money. Then the time slice of the current thread may expire after the check account.getBalance() >= amount passes, but before the withdraw method is called. If, in the interim, another thread withdraws more money, then the test was useless, and we still have a negative balance.

Clearly, the test should be moved inside the withdraw method. That ensures that the test for sufficient funds and the actual withdrawal cannot be separated. Thus, the withdraw method could look like this:

public void withdraw(double amount)
{
   balanceChangeLock.lock();
   try
   {
      while (balance < amount)
      Wait for the balance to grow
      ...
   }
   finally
   {
      balanceChangeLock.unlock();
   }
}

But how can we wait for the balance to grow? We can't simply call sleep inside the withdraw method. If a thread sleeps after acquiring a lock, it blocks all other threads that want to use the same lock. In particular, no other thread can successfully execute the deposit method. Other threads will call deposit, but they will simply be blocked until the withdraw method exits. But the withdraw method doesn't exit until it has funds available. This is the deadlock situation that we mentioned earlier.

To overcome this problem, we use a condition object. Condition objects allow a thread to temporarily release a lock, so that another thread can proceed, and to regain the lock at a later time.

In the telephone booth analogy, suppose that the coin reservoir of the telephone is completely filled, so that no further calls can be made until a service technician removes the coins. You don't want the person in the booth to go to sleep with the door closed. Instead, think of the person leaving the booth temporarily. That gives another person (hopefully a service technician) a chance to enter the booth.

Each condition object belongs to a specific lock object. You obtain a condition object with the newCondition method of the Lock interface. For example,

public class BankAccount
{
   private Lock balanceChangeLock;
   private Condition sufficientFundsCondition;
   ...
   public BankAccount()
   {
      balanceChangeLock = new ReentrantLock();
      sufficientFundsCondition = balanceChangeLock.newCondition();
      ...
   }
}

It is customary to give the condition object a name that describes the condition that you want to test (such as "sufficient funds"). You need to implement an appropriate test. For as long as the test is not fulfilled, call the await method on the condition object:

public void withdraw(double amount)
{
   balanceChangeLock.lock();
   try
   {
      while (balance < amount)
      sufficientFundsCondition.await();
      ...
   }
   finally
   {
      balanceChangeLock.unlock();
   }
}

When a thread calls await, it is not simply deactivated in the same way as a thread that reaches the end of its time slice. Instead, it is in a blocked state, and it will not be activated by the thread scheduler until it is unblocked. To unblock, another thread must execute the signalAll method on the same condition object. The signalAll method unblocks all threads waiting on the condition. They can then compete with all other threads that are waiting for the lock object. Eventually, one of them will gain access to the lock, and it will exit from the await method.

In our situation, the deposit method calls signalAll:

public void deposit(double amount)
{
   balanceChangeLock.lock();
   try
   {
      ...
      sufficientFundsCondition.signalAll();
   }
   finally
   {
      balanceChangeLock.unlock();
   }
}

The call to signalAll notifies the waiting threads that sufficient funds may beavailable, and that it is worth testing the loop condition again.

In the telephone booth analogy, the thread calling await corresponds to the person who enters the booth and finds that the phone doesn't work. That person then leaves the booth and waits outside, depressed, doing absolutely nothing, even as other people enter and leave the booth. The person knows it is pointless to try again. At some point, a service technician enters the booth, empties the coin reservoir, and shouts a signal. Now all the waiting people stop being depressed and again compete for the telephone booth.

There is also a signal method, which randomly picks just one thread that is waiting on the object and unblocks it. The signal method can be more efficient, but it is useful only if you know that every waiting thread can actually proceed. In general, you don't know that, and signal can lead to deadlocks. For that reason, we recommend that you always call signalAll.

The await method can throw an InterruptedException. The withdraw method propagates that exception, because it has no way of knowing what the thread that calls the withdraw method wants to do if it is interrupted.

With the calls to await and signalAll in the withdraw and deposit methods, we can launch any number of withdrawal and deposit threads without a deadlock. If you run the sample program, you will note that all transactions are carried out without ever reaching a negative balance.

ch20/synch/BankAccountThreadRunner.java

1  /**
 2     This program runs threads that deposit and withdraw
 3     money from the same bank account.
 4  */
 5  public class BankAccountThreadRunner
 6  {
 7     public static void main(String[] args)
 8     {
 9        BankAccount account = new BankAccount();
10        final double AMOUNT = 100;
11        final int REPETITIONS = 100;
12        final int THREADS = 100;
13
14        for (int i = 1; i <= THREADS; i++)
15        {
16           DepositRunnable d = new DepositRunnable(
17                  account, AMOUNT, REPETITIONS);
18           WithdrawRunnable w = new WithdrawRunnable(
19                  account, AMOUNT, REPETITIONS);
20
21           Thread dt = new Thread(d);
22           Thread wt = new Thread(w);
23
24           dt.start();
25           wt.start();
26        }
27     }
28  }

ch20/synch/BankAccount.java

1  import java.util.concurrent.locks.Condition;
 2  import java.util.concurrent.locks.Lock;
 3  import java.util.concurrent.locks.ReentrantLock;
 4
 5  /**
 6     A bank account has a balance that can be changed by
 7     deposits and withdrawals.
 8  */
 9  public class BankAccount
10  {
11     private double balance;
12     private Lock balanceChangeLock;
13     private Condition sufficientFundsCondition;
14
15     /**
16        Constructs a bank account with a zero balance.
17     */
18     public BankAccount()
19     {
20        balance = 0;
21        balanceChangeLock = new ReentrantLock();
22        sufficientFundsCondition = balanceChangeLock.newCondition();
23     }
24
25     /**
26        Deposits money into the bank account.
27        @param amount the amount to deposit
28     */
29     public void deposit(double amount)
30     {
31         balanceChangeLock.lock();
32         try
33         {
34           System.out.print("Depositing " + amount);
35           double newBalance = balance + amount;
36           System.out.println(", new balance is " + newBalance);
37           balance = newBalance;
38           sufficientFundsCondition.signalAll();
39        }
40        finally
41        {
42            balanceChangeLock.unlock();
43        }
44     }
45
46     /**
47        Withdraws money from the bank account.
48        @param amount the amount to withdraw
49     */
50     public void withdraw(double amount)
51            throws InterruptedException
52     {
53         balanceChangeLock.lock();
54         try
55         {
56           while (balance < amount)
57              sufficientFundsCondition.await();
58           System.out.print("Withdrawing " + amount);

59           double newBalance = balance - amount;
60           System.out.println(", new balance is " + newBalance);
61           balance = newBalance;
62         }
63         finally
64         {
65             balanceChangeLock.unlock();
66         }
67       }
68
69       /**
70          Gets the current balance of the bank account.
71          @return the current balance
72       */
73       public double getBalance()
74       {
75          return balance;
76       }
77    }

Program Run

Depositing 100.0, new balance is 100.0
Withdrawing 100.0, new balance is 0.0
Depositing 100.0, new balance is 100.0
Depositing 100.0, new balance is 200.0
...
Withdrawing 100.0, new balance is 100.0
Depositing 100.0, new balance is 200.0
Withdrawing 100.0, new balance is 100.0
Withdrawing 100.0, new balance is 0.0

Why is the sufficientFundsCondition object an instance variable of the BankAccount class and not a local variable of the withdraw and deposit methods?

public class BankAccount
{
   public synchronized void deposit(double amount)
   {
      System.out.print("Depositing " + amount);
      double newBalance = balance + amount;
      System.out.println(", new balance is " + newBalance);
      balance = newBalance;
   }

   public synchronized void withdraw(double amount)
   {
      ...
   }
   ...
}

public synchronized void withdraw(double amount)
      throws InterruptedException
{
   while (balance < amount)
      wait();
   ...
}

public synchronized void deposit(double amount)
{
   ...
   notifyAll();
}

One popular use for thread programming is animation. A program that displays an animation shows different objects moving or changing in some way as time progresses. This is often achieved by launching one or more threads that compute how parts of the animation change.

You can use the Swing Timer class for simple animations without having to do any thread programming—see Exercise P20.12 for an example. However, more advanced animations are best implemented with threads.

In this section, you will see a particular kind of animation, namely the visualization of the steps of an algorithm. Algorithm animation is an excellent technique for gaining a better understanding of how an algorithm works. Many algorithms can be animated—type "Java algorithm animation" into your favorite web search engine, and you'll find lots of links to web pages with animations of various algorithms.

All algorithm animations have a similar structure. The algorithm runs in a separate thread that periodically updates an image of the current state of the algorithm and then pauses so that the user can view the image. After a short amount of time, the algorithm thread wakes up again and runs to the next point of interest in the algorithm. It then updates the image and pauses again. This sequence is repeated until the algorithm has finished.

Let's take the selection sort algorithm of Chapter 14 as an example. That algorithm sorts an array of values. It first finds the smallest element, by inspecting all elements in the array, and bringing the smallest element to the leftmost position. It then finds the smallest element among the remaining elements and brings it into the second position. It keeps going in that way. As the algorithm progresses, the sorted part of the array grows.

How can you visualize this algorithm? It is useful to show the part of the array that is already sorted in a different color. Also, we want to show how each step of the algorithm inspects another element in the unsorted part. That demonstrates why the selection sort algorithm is so slow—it first inspects all elements of the array, then all but one, and so on. If the array has n elements, the algorithm inspects

elements. To demonstrate that, we mark the currently visited element in red.

Thus, the algorithm state is described by three items:

This state is accessed by two threads: the thread that sorts the array and the thread that paints the frame. We use a lock to synchronize access to the shared state.

Finally, we add a component instance variable to the algorithm class and augment the constructor to set it. That instance variable is needed for repainting the component and finding out the dimensions of the component when drawing the algorithm state.

public class SelectionSorter
{
   private JComponent component;
   ...
   public SelectionSorter(int[] anArray, JComponent aComponent)
   {
      a = anArray;
      sortStateLock = new ReentrantLock();
      component = aComponent;
   }
}

Figure 20.3. A Step in the Animation of the Selection Sort Algorithm

At each point of interest, the algorithm needs to pause so that the user can admire the graphical output. We supply the pause method shown below, and call it at various places in the algorithm. The pause method repaints the component and sleeps for a small delay that is proportional to the number of steps involved.

public void pause(int steps) throws InterruptedException
{
   component.repaint();
   Thread.sleep(steps * DELAY);
}

We add a draw method to the algorithm class that can draw the current state of the data structure, with the items of special interest highlighted. The draw method is specific to the particular algorithm. This draw method draws the array elements as a sequence of sticks in different colors. The already sorted portion is blue, the marked position is red, and the remainder is black (see Figure 3).

public void draw(Graphics2D g2)
{
   sortStateLock.lock();
   try
    {
      int deltaX = component.getWidth() / a.length;
      for (int i = 0; i < a.length; i++)
      {
         if (i == markedPosition)
            g2.setColor(Color.RED);
          else if (i <= alreadySorted)
            g2.setColor(Color.BLUE);
         else
             g2.setColor(Color.BLACK);
         g2.draw(new Line2D.Double(i * deltaX, 0,
               i * deltaX, a[i]));
      }
    }
    finally
    {

sortStateLock.unlock();
   }
}

You need to update the special positions as the algorithm progresses and pause the animation whenever something interesting happens. The pause should be proportional to the number of steps that are being executed. For a sorting algorithm, pause one unit for each visited array element.

Here is the minimumPosition method from Chapter 14, before the animation code is inserted.

public int minimumPosition(int from)
{
   int minPos = from;
   for (int i = from + 1; i < a.length; i++)
      if (a[i] < a[minPos]) minPos = i;
   return minPos;
}

After each iteration of the for loop, update the marked position of the algorithm state; then pause the program. To measure the cost of each step fairly, pause for two units of time, because two array elements were inspected.

public int minimumPosition(int from)
      throws InterruptedException
{
   int minPos = from;
   for (int i = from + 1; i < a.length; i++)
   {
      sortStateLock.lock();
      try
      {
         if (a[i] < a[minPos]) minPos = i;
         markedPosition = i;
      }
      finally
      {
         sortStateLock.unlock();
      }
      pause(2);
   }
   return minPos;
}

The sort method is augmented in the same way. You will find the code at the end of this section. This concludes the modification of the algorithm class. Let us now turn to the component class.

The component's paintComponent method calls the draw method of the algorithm object.

public class SelectionSortComponent extends JComponent
{
   private SelectionSorter sorter;
   ...
   public void paintComponent(Graphics g)
   {
      if (sorter == null) return;
      Graphics2D g2 = (Graphics2D) g;

sorter.draw(g2);
   }
}

The startAnimation method constructs a SelectionSorter object, which supplies a new array and the this reference to the component that displays the sorted values. Then the method constructs a thread that calls the sorter's sort method.

public void startAnimation()
{
   int[] values = ArrayUtil.randomIntArray(30, 300);
   sorter = new SelectionSorter(values, this);

    class AnimationRunnable implements Runnable
    {
      public void run()
      {
         try
         {
            sorter.sort();
         }
         catch (InterruptedException exception)
         {
         }
      }
   }

   Runnable r = new AnimationRunnable();
   Thread t = new Thread(r);
   t.start();
}

The class for the program that displays the animation is at the end of this section. Run the program and the animation starts.

Exercise P20.7 asks you to animate the merge sort algorithm of Chapter 14. If you do that exercise, then start both programs and run them in parallel to see which algorithm is faster. Actually, you may find the result surprising. If you build fair delays into the merge sort animation to account for the copying from and to the temporary array, you will find that it doesn't perform all that well for small arrays. But if you increase the array size, then the advantage of the merge sort algorithm becomes clear.

ch20/animation/SelectionSortViewer.java

1  import java.awt.BorderLayout;
 2  import javax.swing.JButton;
 3  import javax.swing.JFrame;
 4
 5  public class SelectionSortViewer
 6  {
 7     public static void main(String[] args)
 8     {
 9        JFrame frame = new JFrame();
10
11      final int FRAME_WIDTH = 300;
12      final int FRAME_HEIGHT = 400;
13

14        frame.setSize(FRAME_WIDTH, FRAME_HEIGHT);
15        frame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE);
16
17        final SelectionSortComponent component
18              = new SelectionSortComponent();
19        frame.add(component, BorderLayout.CENTER);
20
21        frame.setVisible(true);
22        component.startAnimation();
23     }
24  }

ch20/animation/SelectionSortComponent.java

1  import java.awt.Graphics;
 2  import java.awt.Graphics2D;
 3  import javax.swing.JComponent;
 4
 5  /**
 6     A component that displays the current state of the selection sort algorithm.
 7  */
 8  public class SelectionSortComponent extends JComponent
 9  {
10     private SelectionSorter sorter;
11
12     /**
13       Constructs the component.
14     */
15     public SelectionSortComponent()
16     {
17       int[] values = ArrayUtil.randomIntArray(30, 300);
18       sorter = new SelectionSorter(values, this);
19     }
20
21     public void paintComponent(Graphics g)
22     {
23        Graphics2D g2 = (Graphics2D)g;
24        sorter.draw(g2);
25     }
26
27     /**
28        Starts a new animation thread.
29     */
30     public void startAnimation()
31     {
32         class AnimationRunnable implements Runnable
33         {
34           public void run()
35           {
36              try
37              {
38                 sorter.sort();
39              }
40              catch (InterruptedException exception)
41              {
42              }
43           }
44         }

45
46          Runnable r = new AnimationRunnable();
47          Thread t = new Thread(r);
48          t.start();
49       }
50    }

ch20/animation/SelectionSorter.java

1  import java.awt.Color;
 2  import java.awt.Graphics2D;
 3  import java.awt.geom.Line2D;
 4  import java.util.concurrent.locks.Lock;
 5  import java.util.concurrent.locks.ReentrantLock;
 6  import javax.swing.JComponent;
 7
 8  /**
 9     This class sorts an array, using the selection sort algorithm.
10  */
11  public class SelectionSorter
12  {
13     private static final int DELAY = 100;
14
15     private int[] a;
16     private Lock sortStateLock;
17
18     // The component is repainted when the animation is paused
19     private JComponent component;
20
21     // These instance variables are needed for drawing
22     private int markedPosition = −1;
23     private int alreadySorted = −1;
24
25     /**
26        Constructs a selection sorter.
27        @param anArray the array to sort
28        @param aComponent the component to be repainted when the animation
29        pauses
30     */
31     public SelectionSorter(int[] anArray, JComponent aComponent)
32     {
33        a = anArray;
34        sortStateLock = new ReentrantLock();
35        component = aComponent;
36     }
37
38     /**
39        Sorts the array managed by this selection sorter.
40     */
41     public void sort()
42           throws InterruptedException
43     {
44        for (int i = 0; i < a.length - 1; i++)
45        {
46           int minPos = minimumPosition(i);
47           sortStateLock.lock();
48           try
49           {

50             swap(minPos, i);
 51             // For animation
 52             alreadySorted = i;
 53           }
 54           finally
 55           {
 56              sortStateLock.unlock();
 57           }
 58           pause(2);
 59        }
 60    }
 61
 62    /**
 63       Finds the smallest element in a tail range of the array.
 64       @param from the first position in a to compare
 65       @return the position of the smallest element in the
 66       range a[from] ... a[a.length - 1]
 67    */
 68    private int minimumPosition(int from)
 69         throws InterruptedException
 70    {
 71      int minPos = from;
 72        for (int i = from + 1; i < a.length; i++)
 73        {
 74           sortStateLock.lock();
 75           try
 76           {
 77              if (a[i] < a[minPos]) minPos = i;
 78              // For animation
 79              markedPosition = i;
 80           }
 81           finally
 82           {
 83               sortStateLock.unlock();
 84           }
 85           pause(2);
 86         }
 87        return minPos;
 88      }
 89
 90      /**
 91         Swaps two entries of the array.
 92         @param i the first position to swap
 93         @param j the second position to swap
 94      */
 95      private void swap(int i, int j)
 96      {
 97         int temp = a[i];
 98         a[i] = a[j];
 99         a[j] = temp;
100      }
101
102      /**
103         Draws the current state of the sorting algorithm.
104         @param g2 the graphics context
105      */
106      public void draw(Graphics2D g2)
107      {

108      sortStateLock.lock();
109      try
110      {
111         int deltaX = component.getWidth() / a.length;
112         for (int i = 0; i < a.length; i++)
113         {
114            if (i == markedPosition)
115               g2.setColor(Color.RED);
116            else if (i <= alreadySorted)
117               g2.setColor(Color.BLUE);
118            else
119                g2.setColor(Color.BLACK);
120            g2.draw(new Line2D.Double(i * deltaX, 0,
121                 i * deltaX, a[i]));
122          }
123      }
124      finally
125      {
126          sortStateLock.unlock();
127          }
128       }
129
130       /**
131          Pauses the animation.
132          @param steps the number of steps to pause
133       */
134       public void pause(int steps)
135              throws InterruptedException
136       {
137          component.repaint();
138          Thread.sleep(steps * DELAY);
139       }
140    }

12.Would the animation still work if the startAnimation method simply called sorter.sort() instead of spawning a thread that calls that method?

Random Fact 20.1 Embedded Systems

An embedded system is a computer system that controls a device. The device contains a processor and other hardware and is controlled by a computer program. Unlike a personal computer, which has been designed to be flexible and run many different computer programs, the hardware and software of an embedded system are tailored to a specific device. Computer-controlled devices are becoming increasingly common, ranging from washing machines to medical equipment, automobile engines, and spacecraft.

Several challenges are specific to programming embedded systems. Most importantly, a much higher standard of quality control applies. Vendors are often unconcerned about bugs in personal computer software, because they can always make you install a patch or upgrade to the next version. But in an embedded system, that is not an option. Few consumers would feel comfortable upgrading the software in their washing machines or automobile engines. If you ever handed in a programming assignment that you believed to be correct, only to have the instructor or grader find bugs in it, then you know how hard it is to write software that can reliably do its task for many years without a chance of changing it.

Figure 20.4. The Controller of an Embedded System

Quality standards are especially important in devices whose failure would destroy property or human life—see Random Facts 7.2 and 11.1.

Many personal computer purchasers buy computers that are fast and have a lot of storage, because the investment is paid back over time when many programs are run on the same equipment. But the hardware for an embedded device is not shared—it is dedicated to one device. A separate processor, memory, and so on, are built for every copy of the device (see Figure 4). If it is possible to shave a few pennies off the manufacturing cost of every unit, the savings can add up quickly for devices that are produced in large volumes. Thus, the embedded-system programmer has a much larger economic incentive to conserve resources than the programmer of desktop software. Unfortunately, trying to conserve resources usually makes it harder to write programs that work correctly.

Generally, embedded systems are written in lower-level programming languages to avoid the overhead of a complex run-time system. The Java run-time system, with its safety mechanisms, garbage collector, support for multithreading, and so on, would be too costly to add to every washing machine. However, some devices are now being built with a scaled-down version of Java: the Java 2 Micro Edition. Examples are smart cell phones and onboard computers for automobiles. The Java 2 Micro Edition is a good candidate for devices that are connected to a network and that need to be able to run new applications safely. For example, you can download a program into a Java-enabled cell phone and be assured that it cannot corrupt other parts of the cell phone software.

java.lang.InterruptedException             java.util.Date
java.lang.Object                           java.util.concurrent.locks.Condition
    notify                                    await
    notifyAll                                 signal
    wait                                      signalAll
java.lang.Runnable                         java.util.concurrent.locks.Lock
    run                                       lock
java.lang.Thread                              newCondition
    interrupted                               unlock
    sleep                                 java.util.concurrent.locks.ReentrantLock
    start

GreetingRunnable r1 = new GreetingRunnable("Hello, World!");
GreetingRunnable r2 = new GreetingRunnable("Goodbye, World!");
r1.run();
r2.run();

java WordCount report.txt address.txt Homework.java

address.txt: 1052
Homework.java: 445
report.txt: 2099

java Find Buff report.txt address.txt Homework.java

report.txt: Buffet style lunch will be available at the
address.txt: Buffet, Warren|11801 Trenton Court|Dallas|TX
Homework.java: BufferedReader in;
address.txt: Walters, Winnie|59 Timothy Circle|Buffalo|MI

public void mergeSort(int from, int to)
{
   if (from == to) return;
   int mid = (from + to) / 2;
   mergeSort(from, mid);
   mergeSort(mid + 1, to);
   merge(from, mid, to);
}