- Java Agents, Applets, and Applications
- Java Standalone Applications
- Java Applications and Multithreading
- The Session Class
- Java Agents, Servlets, Applets, and Applications
- Summary
- Chapter Review
Java Applications and Multithreading
Multithreading is a method of concurrent programming. Multithreading allows two or more programs to work at the same time, cooperatively. When run on a computer with only a single CPU, the system divides the CPU's time among the threads that are ready to run. By sharing the processor time, programs can be written to improve response time or provide more throughput for a given CPU.
In Java, multithreading is built into the language. This feature is one of the reasons why you might want to consider Java rather than LotusScript for certain agents. Some types of programs lend themselves more easily to multithreading than others. If multithreading is one of the requirements of your application, Java is a good choice.
The Java language provides a built-in class called, remarkably enough, Thread. The fully qualified name of this class is java.lang.Thread, so it exists in the java.lang package. Notes provides an important extension to this class called NotesThread. Let's take a look at both of these classes. I'll first describe Thread and then the extensions provided by NotesThread.
The Thread class is used to implement all threads in Java. You can implement your thread in two ways using the Thread class (there is one additional way with NotesThread). The first method is to write your own class that extends the Thread class. In this case, your class actually is a special case of a thread. You write your own class method called run() that overrides the base class method.
The second way to implement a thread involves two separate classes. With this approach, you create your own class that does not extend the Thread class. Instead, you implement an interface called the Runnable interface. This interface defines only one method, the run() method. You create an instance of your class, and then create an instance of the Thread class and pass your class to the Thread constructor. At the appropriate time, your run() method will be called.
You now know that the code of your thread will exist within the run method. When does the code in the run method get invoked? You might think that it would start up as soon as your object is created, but it does not. If that were to happen, you wouldn't get much of a chance to initialize or set things up for your thread before it got off and running. So the creation of your thread and the running of it are separated. To start the code found in your run() method, you invoke the start() method of the thread. This gives you a chance to do some initialization before you invoke the start method. It also means that if you forget to invoke start, your thread will never run.
The start method is located in the Thread class. If your class extends the Thread class, then the start method will be a part of your own class because it is inherited from Thread. If you implemented the Runnable interface, you invoke start from the Thread object, not your own object.
Here is an application showing the first method of implementing threads:
public class MyMain1 { public static void main(String args[ ]) { try { ExtendedThread etMyThread = new ExtendedThread(); // My Thread etMyThread.start(); // This will eventually invoke my run(); etMyThread.join(); // Wait for thread to finish } catch(InterruptedException e) { } } } public class ExtendedThread extends Thread { public void run() { System.out.println("ExtendedThread Running!"); } }
NOTE
I've shown the main Java program and the ExtendedThread classes as separate classes to make it easier to understand. It would be typical in Java to combine these two routines into the same class instead of two different classes. This is because for the javac compiler each separate class must be stored in a separate file with the class name as the filename. If you want to compile with the javac compiler, you'll need to separate the classes into separate files before compiling.
In this example, the ExtendedThread object is created with the new statement. The thread does not actually start until the start method is called. At that time, the start method initializes the thread and invokes its run method. Meanwhile, the original thread returns from the call to start and continues. At this point, it executes the join method, which causes the main thread to wait until the second thread has finished execution. When it does, the main thread wakes up again and finishes the program.
Here is an example of the second method of implementing threads:
public class MyMain2 { public static void main(String args[ ]) { try { RunnableThread rtMyThread = new RunnableThread(); // My Thread Thread theThread = new Thread(rtMyThread); // java.lang.Thread theThread.start(); // This will eventually invoke my run(); theThread.join(); // Wait for thread to finish } catch(InterruptedException e) { } } } class RunnableThread implements Runnable { public void run() { System.out.println("RunnableThread Running!"); } }
In this second example, note that the class RunnableThread is not really a Thread class object; it just runs on a separate thread from the main thread. You can think of the run method in this case as if it were a main program entry point.
In both of these examples, the join method is used to synchronize the main thread with the newly created thread. It would normally be considered a good programming practice for the main thread to wait until the subsidiary threads have finished before it finishes itself.
You might be asking when you should extend the Thread class and when you should use the Runnable interface. If you have complete control over all your classes, it doesn't matter too much which method you use. You can use either. Sometimes, however, you don't have complete control because you might be using some existing classes. In this case, you might not be able to take a class and have it extend the Thread class because you can use only single inheritance in Java, and the class you are working with might already be extending something else. In this situation, you should use the Runnable interface on your existing class because you will then not need to extend the Thread class. You need only to add the run method. Here is an example of the declaration:
public class MyClass extends MyBaseClass implements Runnable { // Here is your MyClass stuff public void run() { // Thread code here } }
In this case, you don't need to modify any code in the MyBaseClass class. In fact, you might not even have the source code for that class. You just need to implement your extensions to the base class, implement one additional method (the run method), and pass your class to the Thread constructor. After that, the base class, as well as your extensions, will be run on a separate thread.
Notes and Domino Multithreading Using Java
You can use the threading mechanisms I described in the previous section within Notes and Domino. However, if you use them, you will not be able to access any Notes or Domino data. You can use the standard Java threading classes only if you are going to use them for animation or other standard Web kinds of processing in, for example, a standard Java applet. If you want to use threading and access Domino databases or services, you must use the NotesThread class.
Now that you've seen the two ways to implement threads with the standard Java libraries, let me show you the modifications necessary to use threads within Notes and Domino. Because NotesThread extends the standard Java Thread class, the first method you can use to create your thread is to extend NotesThread. It works similarly to regular Java, but rather than implementing a run() method, you must implement a runNotes() method. The second way to implement threading with Notes is to implement a Runnable interface. In this case, you still implement a run() method, not a runNotes method. Here are the previous two examples, slightly modified for the Notes and Domino environment. As mentioned, you must supply a runNotes method for this approach to work. In NotesThread, the run method is specified as final, so you will not be allowed to override the run method even if you want to. You will get a compile-time error message if you attempt to use run instead of runNotes. Here is an example of the first method of implementing threads within Notes and Domino:
import lotus.domino.*; public class MyNotesMain1 { public static void main(String args[]) { try { ExtendedThread etMyThread = new ExtendedThread(); // My Thread etMyThread.start(); // This will eventually invoke my runNotes(); etMyThread.join(); // Wait for thread to finish } catch(InterruptedException e) { } } } public class ExtendedThread extends NotesThread { public void runNotes() { System.out.println("ExtendedThread Running!"); } }
Notice that the main program is essentially identical to the version that we created for regular Java threading. The only changes are found in the ExtendedThread class, where it extends NotesThread instead of Thread and contains a runNotes method instead of a run method.
NOTE
You might see some Domino examples and documentation combine the two classes I've shown previously into a single class. This single class extends NotesThread (as in the ExtendedThread) and also has a static main method as well.
I use two classes to illustrate that the main method is called on the caller's thread. The runNotes method is running on a newly created, separate thread. If you combine both the main and runNotes methods in one class, make sure you understand that main and runNotes will be running on separate threads, even though they are defined in the same class.
The same note applies to combining the two classes in the following example.
Here is an example of the second method of implementing threads within Notes and Domino:
import lotus.domino.*; public class MyNotesMain2 { public static void main(String args[ ]) { try { RunnableThread rtMyThread = new RunnableThread(); // My Thread // NotesThread extends java.lang.Thread NotesThread theThread = new NotesThread(rtMyThread); theThread.start(); // This will eventually invoke my run(); theThread.join(); // Wait for thread to finish } catch(InterruptedException e) { } } } class RunnableThread implements Runnable { public void run() { System.out.println("RunnableThread Running!"); } }
You can easily see by comparing the Notes/Domino version with the previous version that the interfaces are very similar. In the Notes version of the applications, it is even more critical for the main routine to wait until the secondary threads have finished (via the join method). This is because Notes or Domino will automatically terminate all the subsidiary threads when the main thread finishes. So the subsidiary threads cannot run longer than the main thread.
It should be apparent from the examples why the NotesThread version uses runNotes and the Runnable version uses the run method. When you extend NotesThread, your class is a NotesThread, which in turn is a Thread class. In the NotesThread class, Lotus has implemented the run method, which will be invoked by Java. The Lotus code within run eventually calls your runNotes method after it initializes the Notes environment. On the other hand, when you implement the Runnable interface, your code is in a separate class, and your run method does not conflict with the Lotus version of the run method. By the time your run method is called, the NotesThread class has already set up the Notes environment. Here is the sequence of calls upon startup of a Runnable class:
-
java.lang.Thread:start initializes the new thread environment and calls the thread's run method on the new thread.
-
NotesThread:run initializes Notes environment and calls the run method of your Runnable class. This method call is made on the NotesThread's thread.
-
YourClass:run is your thread code in the run method.
There is a third way to create a NotesThread in addition to the two that are normally available for all threads. This approach uses static methods within the NotesThread class to initialize Notes on the currently running thread. You use this approach if your program is invoked on a thread over which you have no control. In essence, rather than create a new NotesThread object, this approach initializes Notes without creating a separate NotesThread and turns the currently running thread into a pseudo-NotesThread.
This approach to initializing Notes can also be used if you write a standalone Java application that will use Domino Objects classes. The static initialization might also be required when you are writing event handlers for Java events. Sometimes your methods will be invoked on one of Java's runtime threads (not your own). If you want to call the Domino Objects on your event handler's thread, you will need to statically initialize and terminate during the event handler's operation.
If you do need to initialize Notes on the currently running thread, first you must call NotesThread.sinitThread() before using any other Notes/Domino classes. When you are finished with the Notes classes, you must call NotesThread.stermThread(). These two calls must be balanced. In particular, even if exceptions occur, you must be sure to call stermThread before your routine exits. Here is some sample code:
import lotus.domino.*; public class MyNotesMain3 { public static void main(String args[ ]) { try { // Initialize Notes on the currently running thread NotesThread.sinitThread(); // Here you can place any Java code } catch(Exception e) { e.printStackTrace(); // Print a stack trace } finally { NotesThread.stermThread(); // This will be called in all cases } } }
Notice that in this code, I used the finally clause to house the stermThread call. By implementing the termination this way, the thread Notes environment will be terminated whether or not there is an exception. If you fail to properly terminate the Notes environment, your program can hang or cause an abnormal termination.