posts - 12,  comments - 2,  trackbacks - 0

          JDK6 RC已經發布了,而對于JDK5的新特性還來不及使用,雖然在項目中還沒有使用,但可以寫一些Demo體驗一下Tiger的魅力,現在的時代就是體驗的時代,事事都要親歷親為才能有發言權,怎么有點毛主席“實事求是”的感覺。

          JDK5中的一個亮點就是將Doug Lea并發庫引入到Java標準庫中。Doug Lea確實是一個牛人,能教書,能出書,能編碼,不過這在國外還是比較普遍的,而國內的教授們就相差太遠了。

          一般的服務器都需要線程池,比如Web、FTP等服務器,不過它們一般都自己實現了線程池,比如以前介紹過的Tomcat、Resin和Jetty等,現在有了JDK5,我們就沒有必要重復造車輪了,直接使用就可以,何況使用也很方便,性能也非常高。

          package concurrent;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          public class TestThreadPool {
            public static void main(String args[]) throws InterruptedException {
              // only two threads
              ExecutorService exec = Executors.newFixedThreadPool(2);
              for(int index = 0; index < 100; index++) {
                Runnable run = new Runnable() {
                  public void run() {
                    long time = (long) (Math.random() 1000);
                    System.out.println("Sleeping " + time + "ms");
                      try {
                        Thread.sleep(time);
                      catch (InterruptedException e) {
                      }
                  }
                };
                exec.execute(run);
              }
              // must shutdown
              exec.shutdown();
            }
          }

          上 面是一個簡單的例子,使用了2個大小的線程池來處理100個線程。但有一個問題:在for循環的過程中,會等待線程池有空閑的線程,所以主線程會阻塞的。 為了解決這個問題,一般啟動一個線程來做for循環,就是為了避免由于線程池滿了造成主線程阻塞。不過在這里我沒有這樣處理。[重要修正:經過測試,即使 線程池大小小于實際線程數大小,線程池也不會阻塞的,這與Tomcat的線程池不同,它將Runnable實例放到一個“無限”的 BlockingQueue中,所以就不用一個線程啟動for循環,Doug Lea果然厲害]

          另 外它使用了Executors的靜態函數生成一個固定的線程池,顧名思義,線程池的線程是不會釋放的,即使它是Idle。這就會產生性能問題,比如如果線 程池的大小為200,當全部使用完畢后,所有的線程會繼續留在池中,相應的內存和線程切換(while(true)+sleep循環)都會增加。如果要避 免這個問題,就必須直接使用ThreadPoolExecutor()來構造。可以像Tomcat的線程池一樣設置“最大線程數”、“最小線程數”和“空 閑線程keepAlive的時間”。通過這些可以基本上替換Tomcat的線程池實現方案。

          需要注意的是線程池必須使用shutdown來顯式關閉,否則主線程就無法退出。shutdown也不會阻塞主線程。

           

          多 長時間運行的應用有時候需要定時運行任務完成一些諸如統計、優化等工作,比如在電信行業中處理用戶話單時,需要每隔1分鐘處理話單;網站每天凌晨統計用戶 訪問量、用戶數;大型超時凌晨3點統計當天銷售額、以及最熱賣的商品;每周日進行數據庫備份;公司每個月的10號計算工資并進行轉帳等,這些都是定時任 務。通過 java的并發庫concurrent可以輕松的完成這些任務,而且非常的簡單。

          package concurrent;
          import static java.util.concurrent.TimeUnit.SECONDS;
          import java.util.Date;
          import java.util.concurrent.Executors;
          import java.util.concurrent.ScheduledExecutorService;
          import java.util.concurrent.ScheduledFuture;
          public class TestScheduledThread {
            public static void main(String[] args) {
              final ScheduledExecutorService scheduler = Executors
                  .newScheduledThreadPool(2);
              final Runnable beeper = new Runnable() {
                int count = 0;
                public void run() {
                  System.out.println(new Date() " beep " (++count));
                }
              };
              // 1秒鐘后運行,并每隔2秒運行一次
              final ScheduledFuture<?> beeperHandle = scheduler.scheduleAtFixedRate(
                  beeper, 12, SECONDS);
              // 2秒鐘后運行,并每次在上次任務運行完后等待5秒后重新運行
              final ScheduledFuture<?> beeperHandle2 = scheduler
                  .scheduleWithFixedDelay(beeper, 25, SECONDS);
              // 30秒后結束關閉任務,并且關閉Scheduler
              scheduler.schedule(new Runnable() {
                public void run() {
                  beeperHandle.cancel(true);
                  beeperHandle2.cancel(true);
                  scheduler.shutdown();
                }
              }30, SECONDS);
            }
          }

          為了退出進程,上面的代碼中加入了關閉Scheduler的操作。而對于24小時運行的應用而言,是沒有必要關閉Scheduler的。

           

          實際應用中,有時候需要多個線程同時工作以完成同一件事情,而且在完成過程中,往往會等待其他線程都完成某一階段后再執行,等所有線程都到達某一個階段后再統一執行。

          比如有幾個旅行團需要途經深圳、廣州、韶關、長沙最后到達武漢。旅行團中有自駕游的,有徒步的,有乘坐旅游大巴的;這些旅行團同時出發,并且每到一個目的地,都要等待其他旅行團到達此地后再同時出發,直到都到達終點站武漢。

          這時候CyclicBarrier就可以派上用場。CyclicBarrier最重要的屬性就是參與者個數,另外最要方法是await()。當所有線程都調用了await()后,就表示這些線程都可以繼續執行,否則就會等待。

          package concurrent;
          import java.text.SimpleDateFormat;
          import java.util.Date;
          import java.util.concurrent.BrokenBarrierException;
          import java.util.concurrent.CyclicBarrier;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          public class TestCyclicBarrier {
            // 徒步需要的時間: Shenzhen, Guangzhou, Shaoguan, Changsha, Wuhan
            private static int[] timeWalk = 58151510 };
            // 自駕游
            private static int[] timeSelf = 1344};
            // 旅游大巴
            private static int[] timeBus = 2466};
            
            static String now() {
              SimpleDateFormat sdf = new SimpleDateFormat("HH:mm:ss");
              return sdf.format(new Date()) ": ";
            }

            static class Tour implements Runnable {
              private int[] times;
              private CyclicBarrier barrier;
              private String tourName;
              public Tour(CyclicBarrier barrier, String tourName, int[] times) {
                this.times = times;
                this.tourName = tourName;
                this.barrier = barrier;
              }
              public void run() {
                try {
                  Thread.sleep(times[01000);
                  System.out.println(now() + tourName + " Reached Shenzhen");
                  barrier.await();
                  Thread.sleep(times[11000);
                  System.out.println(now() + tourName + " Reached Guangzhou");
                  barrier.await();
                  Thread.sleep(times[21000);
                  System.out.println(now() + tourName + " Reached Shaoguan");
                  barrier.await();
                  Thread.sleep(times[31000);
                  System.out.println(now() + tourName + " Reached Changsha");
                  barrier.await();
                  Thread.sleep(times[41000);
                  System.out.println(now() + tourName + " Reached Wuhan");
                  barrier.await();
                catch (InterruptedException e) {
                catch (BrokenBarrierException e) {
                }
              }
            }

            public static void main(String[] args) {
              // 三個旅行團
              CyclicBarrier barrier = new CyclicBarrier(3);
              ExecutorService exec = Executors.newFixedThreadPool(3);
              exec.submit(new Tour(barrier, "WalkTour", timeWalk));
              exec.submit(new Tour(barrier, "SelfTour", timeSelf));
              exec.submit(new Tour(barrier, "BusTour", timeBus));
              exec.shutdown();
            }
          }

          運行結果:
          00:02:25: SelfTour Reached Shenzhen
          00:02:25: BusTour Reached Shenzhen
          00:02:27: WalkTour Reached Shenzhen
          00:02:30: SelfTour Reached Guangzhou
          00:02:31: BusTour Reached Guangzhou
          00:02:35: WalkTour Reached Guangzhou
          00:02:39: SelfTour Reached Shaoguan
          00:02:41: BusTour Reached Shaoguan

           

          發庫中的BlockingQueue是一個比較好玩的類,顧名思義,就是阻塞隊列。該類主要提供了兩個方法put()和take(),前者將一個對象放到隊列中,如果隊列已經滿了,就等待直到有空閑節點;后者從head取一個對象,如果沒有對象,就等待直到有可取的對象。

          下 面的例子比較簡單,一個讀線程,用于將要處理的文件對象添加到阻塞隊列中,另外四個寫線程用于取出文件對象,為了模擬寫操作耗時長的特點,特讓線程睡眠一 段隨機長度的時間。另外,該Demo也使用到了線程池和原子整型(AtomicInteger),AtomicInteger可以在并發情況下達到原子化 更新,避免使用了synchronized,而且性能非常高。由于阻塞隊列的put和take操作會阻塞,為了使線程退出,特在隊列中添加了一個“標 識”,算法中也叫“哨兵”,當發現這個哨兵后,寫線程就退出。

          當然線程池也要顯式退出了。

          package concurrent;
          import java.io.File;
          import java.io.FileFilter;
          import java.util.concurrent.BlockingQueue;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          import java.util.concurrent.LinkedBlockingQueue;
          import java.util.concurrent.atomic.AtomicInteger;

          public class TestBlockingQueue {
            static long randomTime() {
              return (long) (Math.random() 1000);
            }

            public static void main(String[] args) {
              // 能容納100個文件
              final BlockingQueue<File> queue = new LinkedBlockingQueue<File>(100);
              // 線程池
              final ExecutorService exec = Executors.newFixedThreadPool(5);
              final File root = new File("F:\\JavaLib");
              // 完成標志
              final File exitFile = new File("");
              // 讀個數
              final AtomicInteger rc = new AtomicInteger();
              // 寫個數
              final AtomicInteger wc = new AtomicInteger();
              // 讀線程
              Runnable read = new Runnable() {
                public void run() {
                  scanFile(root);
                  scanFile(exitFile);
                }

                public void scanFile(File file) {
                  if (file.isDirectory()) {
                    File[] files = file.listFiles(new FileFilter() {
                      public boolean accept(File pathname) {
                        return pathname.isDirectory()
                            || pathname.getPath().endsWith(".java");
                      }
                    });
                    for (File one : files)
                      scanFile(one);
                  else {
                    try {
                      int index = rc.incrementAndGet();
                      System.out.println("Read0: " + index + " "
                          + file.getPath());
                      queue.put(file);
                    catch (InterruptedException e) {
                    }
                  }
                }
              };
              exec.submit(read);
              // 四個寫線程
              for (int index = 0; index < 4; index++) {
                // write thread
                final int NO = index;
                Runnable write = new Runnable() {
                  String threadName = "Write" + NO;
                  public void run() {
                    while (true) {
                      try {
                        Thread.sleep(randomTime());
                        int index = wc.incrementAndGet();
                        File file = queue.take();
                        // 隊列已經無對象
                        if (file == exitFile) {
                          // 再次添加"標志",以讓其他線程正常退出
                          queue.put(exitFile);
                          break;
                        }
                        System.out.println(threadName + ": " + index + " "
                            + file.getPath());
                      catch (InterruptedException e) {
                      }
                    }
                  }
                };
                exec.submit(write);
              }
              exec.shutdown();
            }
          }

          名字可以看出,CountDownLatch是一個倒數計數的鎖,當倒數到0時觸發事件,也就是開鎖,其他人就可以進入了。在一些應用場合中,需要等待某個條件達到要求后才能做后面的事情;同時當線程都完成后也會觸發事件,以便進行后面的操作。

          CountDownLatch最重要的方法是countDown()和await(),前者主要是倒數一次,后者是等待倒數到0,如果沒有到達0,就只有阻塞等待了。

          一個CountDouwnLatch實例是不能重復使用的,也就是說它是一次性的,鎖一經被打開就不能再關閉使用了,如果想重復使用,請考慮使用CyclicBarrier

          下面的例子簡單的說明了CountDownLatch的使用方法,模擬了100米賽跑,10名選手已經準備就緒,只等裁判一聲令下。當所有人都到達終點時,比賽結束。

          同樣,線程池需要顯式shutdown。

          package concurrent;

          import java.util.concurrent.CountDownLatch;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;

          public class TestCountDownLatch {
            public static void main(String[] argsthrows InterruptedException {
              // 開始的倒數鎖
              final CountDownLatch begin = new CountDownLatch(1);
              // 結束的倒數鎖
              final CountDownLatch end = new CountDownLatch(10);
              // 十名選手
              final ExecutorService exec = Executors.newFixedThreadPool(10);
              for(int index = 0; index < 10; index++) {
                final int NO = index + 1;
                Runnable run = new Runnable(){
                  public void run() {
                    try {
                      begin.await();
                      Thread.sleep((long) (Math.random() 10000));
                      System.out.println("No." + NO + " arrived");
                    catch (InterruptedException e) {
                    finally {
                      end.countDown();
                    }
                  }
                };
                exec.submit(run);
              }
              System.out.println("Game Start");
              begin.countDown();
              end.await();
              System.out.println("Game Over");
              exec.shutdown();
            }
          }

          運行結果:
          Game Start
          No.4 arrived
          No.1 arrived
          No.7 arrived
          No.9 arrived
          No.3 arrived
          No.2 arrived
          No.8 arrived
          No.10 arrived
          No.6 arrived
          No.5 arrived
          Game Over

           

          時 候在實際應用中,某些操作很耗時,但又不是不可或缺的步驟。比如用網頁瀏覽器瀏覽新聞時,最重要的是要顯示文字內容,至于與新聞相匹配的圖片就沒有那么重 要的,所以此時首先保證文字信息先顯示,而圖片信息會后顯示,但又不能不顯示,由于下載圖片是一個耗時的操作,所以必須一開始就得下載。

          Java的并發庫Future類 就可以滿足這個要求。Future的重要方法包括get()和cancel(),get()獲取數據對象,如果數據沒有加載,就會阻塞直到取到數據,而 cancel()是取消數據加載。另外一個get(timeout)操作,表示如果在timeout時間內沒有取到就失敗返回,而不再阻塞。

          下面的Demo簡單的說明了Future的使用方法:一個非常耗時的操作必須一開始啟動,但又不能一直等待;其他重要的事情又必須做,等完成后,就可以做不重要的事情。

          package concurrent;

          import java.util.concurrent.Callable;
          import java.util.concurrent.ExecutionException;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          import java.util.concurrent.Future;

          public class TestFutureTask {
            public static void main(String[] args)throws InterruptedException,
                ExecutionException {
              final ExecutorService exec = Executors.newFixedThreadPool(5);
              Callable<String> call = new Callable<String>() {
                public String call() throws Exception {
                  Thread.sleep(1000 5);
                  return "Other less important but longtime things.";
                }
              };
              Future<String> task = exec.submit(call);
              // 重要的事情
              Thread.sleep(1000 3);
              System.out.println("Let's do important things.");
              // 其他不重要的事情
              String obj = task.get();
              System.out.println(obj);
              // 關閉線程池
              exec.shutdown();
            }
          }

          運行結果:
          Let's do important things.
          Other less important but longtime things.

           

          慮以下場景:瀏覽網頁時,瀏覽器了5個線程下載網頁中的圖片文件,由于圖片大小、網站訪問速度等諸多因素的影響,完成圖片下載的時間就會有很大的不同。如果先下載完成的圖片就會被先顯示到界面上,反之,后下載的圖片就后顯示。

          Java的并發庫CompletionService可 以滿足這種場景要求。該接口有兩個重要方法:submit()和take()。submit用于提交一個runnable或者callable,一般會提 交給一個線程池處理;而take就是取出已經執行完畢runnable或者callable實例的Future對象,如果沒有滿足要求的,就等待了。 CompletionService還有一個對應的方法poll,該方法與take類似,只是不會等待,如果沒有滿足要求,就返回null對象。

          package concurrent;

          import java.util.concurrent.Callable;
          import java.util.concurrent.CompletionService;
          import java.util.concurrent.ExecutionException;
          import java.util.concurrent.ExecutorCompletionService;
          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          import java.util.concurrent.Future;

          public class TestCompletionService {
            public static void main(String[] argsthrows InterruptedException,
                ExecutionException {
              ExecutorService exec = Executors.newFixedThreadPool(10);
              CompletionService<String> serv = 
                new ExecutorCompletionService<String>(exec);

              for (int index = 0; index < 5; index++) {
                final int NO = index;
                Callable<String> downImg = new Callable<String>() {
                  public String call() throws Exception {
                    Thread.sleep((long) (Math.random() 10000));
                    return "Downloaded Image " + NO;
                  }
                };
                serv.submit(downImg);
              }

              Thread.sleep(1000 2);
              System.out.println("Show web content");
              for (int index = 0; index < 5; index++) {
                Future<String> task = serv.take();
                String img = task.get();
                System.out.println(img);
              }
              System.out.println("End");
              // 關閉線程池
              exec.shutdown();
            }
          }

          運行結果:
          Show web content
          Downloaded Image 1
          Downloaded Image 2
          Downloaded Image 4
          Downloaded Image 0
          Downloaded Image 3
          End

           

          操作系統的信號量是個很重要的概念,在進程控制方面都有應用。Java并發庫Semaphore可以很輕松完成信號量控制,Semaphore可以控制某個資源可被同時訪問的個數,acquire()獲取一個許可,如果沒有就等待,而release()釋放一個許可。比如在Windows下可以設置共享文件的最大客戶端訪問個數。

          Semaphore維護了當前訪問的個數,提供同步機制,控制同時訪問的個數。在數據結構中鏈表可以保存“無限”的節點,用Semaphore可以實現有限大小的鏈表。另外重入鎖ReentrantLock也可以實現該功能,但實現上要負責些,代碼也要復雜些。

          下面的Demo中申明了一個只有5個許可的Semaphore,而有20個線程要訪問這個資源,通過acquire()和release()獲取和釋放訪問許可。

          package concurrent;

          import java.util.concurrent.ExecutorService;
          import java.util.concurrent.Executors;
          import java.util.concurrent.Semaphore;

          public class TestSemaphore {
            public static void main(String[] args) {
              // 線程池
              ExecutorService exec = Executors.newCachedThreadPool();
              // 只能5個線程同時訪問
              final Semaphore semp = new Semaphore(5);
              // 模擬20個客戶端訪問
              for (int index = 0; index < 20; index++) {
                final int NO = index;
                Runnable run = new Runnable() {
                  public void run() {
                    try {
                      // 獲取許可
                      semp.acquire();
                      System.out.println("Accessing: " + NO);
                      Thread.sleep((long) (Math.random() 10000));
                      // 訪問完后,釋放
                      semp.release();
                    catch (InterruptedException e) {
                    }
                  }
                };
                exec.execute(run);
              }
              // 退出線程池
              exec.shutdown();
            }
          }

          運行結果:
          Accessing: 0
          Accessing: 1
          Accessing: 2
          Accessing: 3
          Accessing: 4
          Accessing: 5
          Accessing: 6
          Accessing: 7
          Accessing: 8
          Accessing: 9
          Accessing: 10
          Accessing: 11
          Accessing: 12
          Accessing: 13
          Accessing: 14
          Accessing: 15
          Accessing: 16
          Accessing: 17
          Accessing: 18
          Accessing: 19

          posted on 2008-08-06 14:27 doodoosun 閱讀(165) 評論(0)  編輯  收藏 所屬分類: Java 相關

          <2025年5月>
          27282930123
          45678910
          11121314151617
          18192021222324
          25262728293031
          1234567

          常用鏈接

          留言簿(1)

          隨筆分類(6)

          隨筆檔案(11)

          文章分類(30)

          文章檔案(34)

          搜索

          •  

          最新評論

          閱讀排行榜

          評論排行榜

          主站蜘蛛池模板: 六安市| 洛浦县| 渭南市| 兴安盟| 泰来县| 汝阳县| 乌拉特后旗| 化州市| 普定县| 乌海市| 门头沟区| 大竹县| 嘉兴市| 卢湾区| 酒泉市| 黎平县| 赤壁市| 礼泉县| 淮阳县| 松阳县| 丘北县| 恭城| 镇宁| 凉城县| 九台市| 海南省| 北流市| 黎平县| 宾川县| 德昌县| 隆昌县| 会东县| 潍坊市| 宜阳县| 邵阳市| 九江市| 滦平县| 离岛区| 无锡市| 彰武县| 综艺|