Android开发经验谈：并发编程（线程与线程池）（推荐）

一、线程

在Android开发中，你不可能都在主线程中开发，毕竟要联网，下载数据，保存数据等操作，当然这就离不开线程。

（当然你可以在Android4.0以前的手机里在主线程请求网络，我最早开发的时候，用的手机比较古老。。。）

在Android中你可以随意创建线程，于是就会造成线程不可控，内存泄漏，创建线程消耗资源，线程太多了消耗资源等问题。
具体线程怎么创建我就不在文章里描述了，毕竟这主要将并发编程。。。。

大家知道线程不可控就好了。。。于是就需要对线程进行控制，防止一系列问题出现，这就用到了如下要讲的东西。

二、线程池

线程池：顾名思义，就是放线程的大池子。

如何创建一个线程池？

先说说几个系统的线程池：

FixedThreadPool 创建定长线程的线程池
CachedThreadPool 需要的时候建立新的线程，超时线程销毁
SingleThreadPool 单个线程的线程池
ScheduledThreadPool 可以定时的线程池，创建周期性的任务

这几个线程池不做多余阐述，因为这些线程池的原理都与我下面要讲的有关。。。。

如何自定义线程池（先来了解几个必须知道的参数）：

corePoolSize：

核心线程池大小，线程池中主要工作的线程的多少。

maximumPoolSize：

线程池最大线程数。

keepAliveTime：

空闲线程可保持的时间是多久，如果你启用了allowCoreThreadTimeOut方法，你的线程池里的空闲线程在这个时间段后会自动销毁，如果没启用，则只要不超过corePoolSize，空闲线程也不会销毁。

Unit：

keepAliveTime的时间单位

workQueue：

阻塞队列，当任务达到corePoolSize，就会被放入这个队列

常见几种BlockingQueue实现

ArrayBlockingQueue : 有界的数组队列
LinkedBlockingQueue : 可支持有界/无界的队列，使用链表实现
PriorityBlockingQueue : 优先队列，可以针对任务排序
SynchronousQueue : 队列长度为1的队列，和Array有点区别就是：client thread提交到block queue会是一个阻塞过程，直到有一个worker thread连接上来poll task。

threadFactory：

线程工厂，主要用来创建线程；

handler：

表示当拒绝处理任务时的策略，也就是参数maximumPoolSize达到后丢弃处理的方法。有以下四种取值：

ThreadPoolExecutor.AbortPolicy:丢弃任务并抛出RejectedExecutionException异常。
ThreadPoolExecutor.DiscardPolicy：也是丢弃任务，但是不抛出异常。
ThreadPoolExecutor.DiscardOldestPolicy：丢弃队列最前面的任务，然后重新尝试执行任务（重复此过程）
ThreadPoolExecutor.CallerRunsPolicy：由调用线程处理该任务

用户也可以实现接口RejectedExecutionHandler定制自己的策略。

代码展示：

//线程工厂
public class TaskThreadFactory implements ThreadFactory {

 private final AtomicInteger mThreadNumber = new AtomicInteger(1);



 private final String mNamePrefix;

 TaskThreadFactory(String name) {
  mNamePrefix = name + "#";
 }

 public Thread newThread(Runnable r) {
  Thread t = new Thread(r,mNamePrefix + mThreadNumber.getAndIncrement());

//  if (t.isDaemon())
//   t.setDaemon(false);
//
//  if (t.getPriority() != Thread.NORM_PRIORITY)
//   t.setPriority(Thread.NORM_PRIORITY);

  return t;
 }
}

//重写runnable
public class PRunnable implements Runnable {

 public static final int HIGH = 1;//优先级高
 public static final int NORMAL = 2;//优先级中等
 public static final int LOW = 3;//优先级低
 @IntDef({HIGH,NORMAL,LOW})
 @Retention(RetentionPolicy.SOURCE)
 public @interface Priority{}

 public final int priority;
 private final Runnable runnable;
 public int serial;

 public PRunnable(Runnable runnable){
  this(NORMAL,runnable);
 }

 public PRunnable(@Priority int priority,Runnable runnable){
  this.priority = priority;
  this.runnable = runnable;
 }

 @Override
 public void run() {
  if (runnable != null) {
   runnable.run();
  }
 }

 /**
  * 线程队列方式 先进先出
  * @param r1
  * @param r2
  * @return
  */
 public static final int compareFIFO(PRunnable r1, PRunnable r2) {
  int result = r1.priority-r2.priority;
  return result==0?r1.serial-r2.serial:result;
 }

 /**
  * 线程队列方式 后进先出
  * @param r1
  * @param r2
  * @return
  */
 public static final int compareLIFO(PRunnable r1, PRunnable r2) {
  int result = r1.priority-r2.priority;
  return result==0?r2.serial-r1.serial:result;
 }
}

//线程池实现
public class TaskExecutor implements Executor {

 private final static int QUEUE_INIT_CAPACITY = 20;

 private static final int CORE = 3;

 private static final int MAX = 5;

 private static final int TIMEOUT = 30 * 1000;

 private AtomicInteger SERIAL = new AtomicInteger(0);//主要获取添加任务

 public static class Config {
  public int core;

  public int max;

  public int timeout;

  public boolean allowCoreTimeOut;

  public boolean fifo;

  public Config(int core, int max, int timeout, boolean allowCoreTimeOut,boolean fifo) {
   this.core = core;
   this.max = max;
   this.timeout = timeout;
   this.allowCoreTimeOut = allowCoreTimeOut;
   this.fifo = fifo;
  }
 }

 public static Config defaultConfig = new Config(CORE, MAX, TIMEOUT, true,true);


 private final String name;

 private final Config config;

 private ExecutorService service;

 public TaskExecutor(String name) {
  this(name, defaultConfig);
 }

 public TaskExecutor(String name, Config config) {
  this(name, config, true);
 }

 public TaskExecutor(String name, Config config, boolean startup) {
  this.name = name;
  this.config = config;

  if (startup) {
   startup();
  }
 }

 public void startup() {
  synchronized (this) {
   if (service != null && !service.isShutdown()) {
    return;
   }

   service = createExecutor(config);
  }
 }

 public void shutdown() {
  ExecutorService executor = null;

  synchronized (this) {
   // 交换变量
   if (service != null) {
    executor = service;
    service = null;
   }
  }

  if (executor != null) {
   // 停止线程
   if (!executor.isShutdown()) {
    executor.shutdown();
   }

   // 回收变量
   executor = null;
  }
 }

 private void executeRunnable(PRunnable runnable) {
  synchronized (this) {
   if (service == null || service.isShutdown()) {
    return;
   }
   runnable.serial = SERIAL.getAndIncrement();
   service.execute(runnable);
  }
 }

 @Override
 public void execute(Runnable runnable) {
  if (runnable instanceof PRunnable) {
   executeRunnable((PRunnable) runnable);
  }else{
   executeRunnable(new PRunnable(runnable));
  }
 }

 public Future<?> submit(Runnable runnable) {
  synchronized (this) {
   if (service == null || service.isShutdown()) {
    return null;
   }
   if (runnable instanceof PRunnable) {
    ((PRunnable) runnable).serial = SERIAL.getAndIncrement();
    return service.submit(runnable);
   }else{
    PRunnable pRunnable = new PRunnable(runnable);
    pRunnable.serial = SERIAL.getAndIncrement();
    return service.submit(pRunnable);
   }
  }
 }

 public void execute(Runnable runnable, @PRunnable.Priority int priority) {
  executeRunnable(new PRunnable(priority,runnable));
 }

 private ExecutorService createExecutor(Config config) {
  ThreadPoolExecutor service = new ThreadPoolExecutor(config.core, config.max, config.timeout,
    TimeUnit.MILLISECONDS, new PriorityBlockingQueue<Runnable>(QUEUE_INIT_CAPACITY, config.fifo ? mQueueFIFOComparator : mQueueLIFOComparator),
    new TaskThreadFactory(name), new ThreadPoolExecutor.DiscardPolicy());

  allowCoreThreadTimeOut(service, config.allowCoreTimeOut);

  return service;
 }

 public boolean isBusy() {
  synchronized (this) {
   if (service == null || service.isShutdown()) {
    return false;
   }
   if(service instanceof ThreadPoolExecutor){
    ThreadPoolExecutor tService = (ThreadPoolExecutor) service;
    return tService.getActiveCount() >= tService.getCorePoolSize();
   }
   return false;
  }
 }

 private static final void allowCoreThreadTimeOut(ThreadPoolExecutor service, boolean value) {
  if (Build.VERSION.SDK_INT >= 9) {
   allowCoreThreadTimeOut9(service, value);
  }
 }

 @TargetApi(9)
 private static final void allowCoreThreadTimeOut9(ThreadPoolExecutor service, boolean value) {
  service.allowCoreThreadTimeOut(value);
 }


 Comparator<Runnable> mQueueFIFOComparator = new Comparator<Runnable>() {

  @Override
  public int compare(Runnable lhs, Runnable rhs) {
   PRunnable r1 = (PRunnable) lhs;
   PRunnable r2 = (PRunnable) rhs;

   return PRunnable.compareFIFO(r1, r2);
  }
 };

 Comparator<Runnable> mQueueLIFOComparator = new Comparator<Runnable>() {

  @Override
  public int compare(Runnable lhs, Runnable rhs) {
   PRunnable r1 = (PRunnable) lhs;
   PRunnable r2 = (PRunnable) rhs;

   return PRunnable.compareLIFO(r1, r2);
  }
 };

}

频道导航

Android开发经验谈：并发编程（线程与线程池）（推荐）

您可能感兴趣的文章: