Thread.join()方法解析

API：

join

public final void join()                throws InterruptedException

等待该线程终止。

抛出：

InterruptedException - 如果任何线程中断了当前线程。当抛出该异常时，当前线程的中断状态 被清除。

join

public final void join(long millis)                throws InterruptedException

等待该线程终止的时间最长为 millis 毫秒。超时为 0 意味着要一直等下去。

参数：

millis - 以毫秒为单位的等待时间。

抛出：

InterruptedException - 如果任何线程中断了当前线程。当抛出该异常时，当前线程的中断状态 被清除。

join

public final void join(long millis,                       int nanos)                throws InterruptedException

等待该线程终止的时间最长为 millis 毫秒 + nanos 纳秒。

参数：

millis - 以毫秒为单位的等待时间。

nanos - 要等待的 0-999999 附加纳秒。

抛出：

IllegalArgumentException - 如果 millis 值为负，则 nanos 的值不在 0-999999 范围内。

InterruptedException - 如果任何线程中断了当前线程。当抛出该异常时，当前线程的中断状态 被清除。

解析：

Thread.join()，是用来指定当前主线程等待其他线程执行完毕后，再来继续执行Thread.join()后面的代码。

例1：

package com.example;import java.util.Date;import java.util.concurrent.TimeUnit;public class DataSourcesLoader implements Runnable{@Overridepublic void run() {System.out.printf("Beginning data sources loading: %s\n",new Date());    try {      TimeUnit.SECONDS.sleep(4);    } catch (InterruptedException e) {      e.printStackTrace();    }    System.out.printf("Data sources loading has finished: %s\n",new Date());}public static void main(String[] args){DataSourcesLoader dsLoader = new DataSourcesLoader();    Thread thread1 = new Thread(dsLoader,"DataSourceThread");        thread1.start();        try {        thread1.join();      } catch (InterruptedException e) {        e.printStackTrace();      }            System.out.printf("Main: Configuration has been loaded: %s\n",new Date());}}

执行结果：

Beginning data sources loading: Fri Nov 14 14:27:31 CST 2014Data sources loading has finished: Fri Nov 14 14:27:35 CST 2014Main: Configuration has been loaded: Fri Nov 14 14:27:35 CST 2014

如果去掉thread1.join()，执行的结果如下：

Main: Configuration has been loaded: Fri Nov 14 14:28:33 CST 2014Beginning data sources loading: Fri Nov 14 14:28:33 CST 2014Data sources loading has finished: Fri Nov 14 14:28:37 CST 2014

通过结果，就可以很明显的说明上面红字的部分：“再来继续执行Thread.join()后面的代码”

例2：

package com.example;import java.util.Date;import java.util.concurrent.TimeUnit;public class DataSourcesLoader implements Runnable{@Overridepublic void run() {System.out.printf("Beginning data sources loading: %s\n",new Date());    try {      TimeUnit.SECONDS.sleep(4);    } catch (InterruptedException e) {      e.printStackTrace();    }    System.out.printf("Data sources loading has finished: %s\n",new Date());}public static void main(String[] args){DataSourcesLoader dsLoader = new DataSourcesLoader();    Thread thread1 = new Thread(dsLoader,"DataSourceThread");        thread1.start();        try {        thread1.join(3000);      } catch (InterruptedException e) {        e.printStackTrace();      }            System.out.printf("Main: Configuration has been loaded: %s\n",new Date());}}

这里使用的是：

thread1.join(3000);

这句话的意思是，只要满足下面2个条件中的一个时，主线程就会继续执行thread.join()后面的代码：

条件1：thread1 执行完毕；

条件2：已经等待 thread1 执行了3000ms.

例子中，thread1 自身的执行时间是4s，而设置的等待时间是3s，所以得到的执行结果如下，thread1还没有执行完，主线程就开始执行后面的代码，因为 thread1 等待的时间已经超时了：

Beginning data sources loading: Fri Nov 14 14:35:45 CST 2014Main: Configuration has been loaded: Fri Nov 14 14:35:48 CST 2014Data sources loading has finished: Fri Nov 14 14:35:49 CST 2014

那么结合上面的2个例子，我们可以推断出下面代码的执行结果了：

例3：

package com.example;import java.util.Date;import java.util.concurrent.TimeUnit;public class DataSourcesLoader implements Runnable{@Overridepublic void run() {System.out.printf("Beginning data sources loading: %s\n",new Date());    try {      TimeUnit.SECONDS.sleep(4);    } catch (InterruptedException e) {      e.printStackTrace();    }    System.out.printf("Data sources loading has finished: %s\n",new Date());}}

package com.example;import java.util.Date;import java.util.concurrent.TimeUnit;public class NetworkConnectionsLoader implements Runnable{@Overridepublic void run() {System.out.printf("Beginning network connect loading: %s\n",new Date());    try {      TimeUnit.SECONDS.sleep(6);    } catch (InterruptedException e) {      e.printStackTrace();    }    System.out.printf("Network connect loading has finished: %s\n",new Date());}public static void main(String[] args){DataSourcesLoader dsLoader = new DataSourcesLoader();    Thread thread1 = new Thread(dsLoader,"DataSourceThread");        NetworkConnectionsLoader ncLoader = new NetworkConnectionsLoader();    Thread thread2 = new Thread(ncLoader,"NetworkConnectionLoader");        thread1.start();    thread2.start();         try {        thread1.join();        thread2.join(1900);      } catch (InterruptedException e) {        e.printStackTrace();      }            System.out.printf("Main: Configuration has been loaded: %s\n",new Date());}}

执行结果：

Beginning data sources loading: Fri Nov 14 14:39:20 CST 2014Beginning network connect loading: Fri Nov 14 14:39:20 CST 2014Data sources loading has finished: Fri Nov 14 14:39:24 CST 2014Main: Configuration has been loaded: Fri Nov 14 14:39:26 CST 2014Network connect loading has finished: Fri Nov 14 14:39:26 CST 2014

注意：如果把例3的 thread2.join(1900) 部分修改为：

thread2.join(3000);

结果会和上面的一样吗？

根据我最开始指出的“Thread.join()，是用来指定当前主线程等待其他线程执行完毕后，再来继续执行Thread.join()后面的代码。”

我们可以看到，执行结果会有差别：

Beginning data sources loading: Fri Nov 14 14:41:21 CST 2014Beginning network connect loading: Fri Nov 14 14:41:21 CST 2014Data sources loading has finished: Fri Nov 14 14:41:25 CST 2014Network connect loading has finished: Fri Nov 14 14:41:27 CST 2014Main: Configuration has been loaded: Fri Nov 14 14:41:27 CST 2014</span>

至于为什么会有这个差别，我上面也已经说明了，我想这个应该不难理解。

PS：代码部分截取来自《Java 7 Concurrency Cookbook》