Trail: Essential Classes
Lesson: Concurrency
Home Page > Essential Classes > Concurrency
Answers to Questions and Exercises: Concurrency

Questions

  1. Question: Can you pass a Thread object to Executor.execute? Would such an invocation make sense? Why or why not?

    Answer: Thread implements the Runnable interface, so you can pass an instance of Thread to Executor.execute. However it doesn't make sense to use Thread objects this way. If the object is directly instantiated from Thread, its run method doesn't do anything. You can define a subclass of Thread with a useful run method — but such a class would implement features that the executor would not use.

Exercises

  1. Exercise: Compile and run BadThreads.java:
    public class BadThreads {
    
        static String message;
    
        private static class CorrectorThread extends Thread {
    
            public void run() {
                try {
                    sleep(1000); 
                } catch (InterruptedException e) {}
                //Key statement 1:
                message = "Mares do eat oats."; 
            }
        }
    
        public static void main(String args[]) throws InterruptedException {
            (new CorrectorThread()).start();
            message = "Mares do not eat oats.";
            Thread.sleep(2000);
            //Key statement 2:
            System.out.println(message);
        }
    }
    

    The application should print out "Mares do eat oats." Is it guaranteed to always do this? If not, why not? Would it help to change the parameters of the two invocations of Sleep? Describe two ways to change the program to enforce such a guarantee.

    Solution: The program will almost always print out "Mares do eat oats." However, this result is not guaranteed, because there is no happens-before relationship between "Key statement 1" and "Key statment 2". This is true even if "Key statement 1" actually executes before "Key statement 2" — remember, a happens-before relationship is about visibility, not sequence.

    There are two ways you can guarantee that all changes to message will be visible to the main thread:

    • In the main thread, retain a reference to the CorrectorThread instance. Then invoke join on that instance before referring to message
    • Encapsulate message in an object with synchronized methods. Never reference message except through those methods.

    Both of these techniques establish the necessary happens-before relationship, making changes to message visible.

    A third technique is to simply declare message as volatile. This guarantees that any write to message (as in "Key statement 1") will have a happens-before relationship with any subsequent reads of message (as in "Key statement 2"). But it does not guarantee that "Key statement 1" will literally happen before "Key statement 2". They will probably happen in sequence, but because of scheduling uncertainities and the unknown granularity of sleep, this is not guaranteed.

    Changing the arguments of the two sleep invocations does not help either, since this does nothing to guarantee a happens-before relationship.

  2. Exercise: Modify the producer-consumer example in Guarded Blocks to use a standard library class instead of the Drop class.

    Solution: The java.util.concurrent.BlockingQueue interface defines a get method that blocks if the queue is empty, and a put methods that blocks if the queue is full. These are effectively the same operations defined by Drop — except that Drop is not a queue! However, there's another way of looking at Drop: it's a queue with a capacity of zero. Since there's no room in the queue for any elements, every get blocks until the corresponding take and every take blocks until the corresponding get. There is an implementation of BlockingQueue with precisely this behavior: java.util.concurrent.SynchronousQueue.

    BlockingQueue is almost a drop-in replacement for Drop. The main problem in Producer is that with BlockingQueue, the put and get methods throw InterruptedException. This means that the existing try must be moved up a level:

    import java.util.Random;
    import java.util.concurrent.BlockingQueue;
    
    public class Producer implements Runnable {
        private BlockingQueue<String> drop;
    
        public Producer(BlockingQueue<String> drop) {
            this.drop = drop;
        }
    
        public void run() {
            String importantInfo[] = {
                "Mares eat oats",
                "Does eat oats",
                "Little lambs eat ivy",
                "A kid will eat ivy too"
            };
            Random random = new Random();
    
            try {
                for (int i = 0; i < importantInfo.length; i++) {
                    drop.put(importantInfo[i]);
                    Thread.sleep(random.nextInt(5000));
                }
                drop.put("DONE");
            } catch (InterruptedException e) {}
        }
    }
    
    
    Similar changes are required for Consumer:
    import java.util.Random;
    import java.util.concurrent.BlockingQueue;
    
    public class Consumer implements Runnable {
        private BlockingQueue<String> drop;
    
        public Consumer(BlockingQueue<String> drop) {
            this.drop = drop;
        }
    
        public void run() {
            Random random = new Random();
            try {
                for (String message = drop.take(); ! message.equals("DONE");
                        message = drop.take()) {
                    System.out.format("MESSAGE RECEIVED: %s%n", message);
                    Thread.sleep(random.nextInt(5000));
                }
            } catch (InterruptedException e) {}
        }
    }
                    
    
    
    For ProducerConsumerExample, we simply change the declaration for the drop object:
    import java.util.concurrent.BlockingQueue;
    import java.util.concurrent.SynchronousQueue;
    
    public class ProducerConsumerExample {
        public static void main(String[] args) {
            BlockingQueue<String> drop = new SynchronousQueue<String> ();
            (new Thread(new Producer(drop))).start();
            (new Thread(new Consumer(drop))).start();
        }
    }
    
Previous page: Questions and Exercises: Concurrency