iOS中多线程编程工具主要有:
NSThread NSOperation GCD 这三种方法都简单易用,各有千秋.但无疑GCD是最有诱惑力的,因为其本身是apple为多核的并行运算提出的解决方案.虽然当前移动平台用双核的不多,但不影响GCD作为多线程编程的利器(ipad2已经是双核了,这无疑是一个趋势).
GCD是和block紧密相连的,所以最好先了解下block(可以查看这里).GCD是C level的函数,这意味着它也提供了C的函数指针作为参数,方便了C程序员.
一、下面首先来看GCD的使用:
dispatch_async(dispatch_queue_t queue, dispatch_block_t block); async表明异步运行,block代表的是你要做的事情,queue则是你把任务交给谁来处理了.(除了async,还有sync,delay,本文以async为例).
之所以程序中会用到多线程是因为程序往往会需要读取数据,然后更新UI.为了良好的用户体验,读取数据的操作会倾向于在后台运行,这样以避免阻塞主线程.GCD里就有三种queue来处理.
先来介绍一下 Main queue:
顾名思义,运行在主线程,由dispatch_get_main_queue获得.和ui相关的就要使用Main Queue.
[cpp] view plaincopy //GCD下载图片刷新主界面的例子
/*
(IBAction)touchUpInsideByThreadOne:(id)sender { dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{ NSURL url = [NSURL URLWithString:@"http://avatar.csdn.net/2/C/D/1_totogo2010.jpg"]; NSData data = [[NSData alloc]initWithContentsOfURL:url]; UIImage image = [[UIImage alloc]initWithData:data]; if (data != nil) { dispatch_async(dispatch_get_main_queue(), ^{ self.imageView.image = image; }); } }); }/
通过与线程池的配合,dispatch queue分为下面两种:而系统默认就有一个串行队列main_queue和并行队列global_queue:
Serial Dispatch Queue -- 线程池只提供一个线程用来执行任务,所以后一个任务必须等到前一个任务执行结束才能开始。 Concurrent Dispatch Queue -- 线程池提供多个线程来执行任务,所以可以按序启动多个任务并发执行。 而系统默认就有一个串行队列main_queue和并行队列global_queue:
[cpp] view plaincopy dispatch_queue_t globalQ = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
dispatch_queue_t mainQ = dispatch_get_main_queue();
通常,我们可以在global_queue中做一些long-running的任务,完成后在main_queue中更新UI,避免UI阻塞,无法响应用户操作: [cpp] view plaincopy dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
// long-running task
dispatch_async(dispatch_get_main_queue(), ^{
// update UI
});
});
1.Serial quque(private dispatch queue)
每次运行一个任务,可以添加多个,执行次序FIFO. 通常是指程序员生成的,比如:
NSDate da = [NSDate date]; NSString daStr = [da description]; const char *queueName = [daStr UTF8String]; dispatch_queue_t myQueue = dispatch_queue_create(queueName, DISPATCH_QUEUE_PRIORITY_DEFAULT); 下面还是下载图片例子:
[cpp] view plaincopy
(IBAction)touchUpInsideByThreadOne:(id)sender {
NSDate da = [NSDate date];
NSString daStr = [da description];
const char *queueName = [daStr UTF8String];
dispatch_queue_t myQueue = dispatch_queue_create(queueName, NULL);
dispatch_async(myQueue, ^{
NSURL url = [NSURL URLWithString:@"http://avatar.csdn.net/2/C/D/1_totogo2010.jpg"];
NSData data = [[NSData alloc]initWithContentsOfURL:url];
UIImage *image = [[UIImage alloc]initWithData:data];
if (data != nil) {
dispatch_async(dispatch_get_main_queue(), ^{
self.imageView.image = image;
});
}
});
dispatch_release(myQueue);
}
为了验证Serial queue的FIFO特性,写了如下的验证代码:发现的确是顺序执行的。
[cpp] view plaincopy
(IBAction)touchUpInsideByThreadOne:(id)sender {
NSDate da = [NSDate date];
NSString daStr = [da description];
const char *queueName = [daStr UTF8String];
dispatch_queue_t myQueue = dispatch_queue_create(queueName, DISPATCH_QUEUE_SERIAL);
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:6];
NSLog(@"[NSThread sleepForTimeInterval:6];");
});
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:3];
NSLog(@"[NSThread sleepForTimeInterval:3];");
});
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:1];
NSLog(@"[NSThread sleepForTimeInterval:1];");
});
dispatch_release(myQueue);
}
运行结果为:
[cpp] view plaincopy 2013-07-24 16:37:14.397 NSThreadAndBlockDemo[1924:12303] [NSThread sleepForTimeInterval:6];
2013-07-24 16:37:17.399 NSThreadAndBlockDemo[1924:12303] [NSThread sleepForTimeInterval:3];
2013-07-24 16:37:18.401 NSThreadAndBlockDemo[1924:12303] [NSThread sleepForTimeInterval:1];
Concurrent queue(global dispatch queue):
可以同时运行多个任务,每个任务的启动时间是按照加入queue的顺序,结束的顺序依赖各自的任务.使用dispatch_get_global_queue获得.
[cpp] view plaincopy
(IBAction)touchUpInsideByThreadOne:(id)sender {
dispatch_queue_t myQueue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:6];
NSLog(@"[NSThread sleepForTimeInterval:6];");
});
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:3];
NSLog(@"[NSThread sleepForTimeInterval:3];");
});
dispatch_async(myQueue, ^{
[NSThread sleepForTimeInterval:1];
NSLog(@"[NSThread sleepForTimeInterval:1];");
});
dispatch_release(myQueue);
}
运行的结果为: [cpp] view plaincopy 2013-07-24 16:38:41.660 NSThreadAndBlockDemo[1944:12e03] [NSThread sleepForTimeInterval:1];
2013-07-24 16:38:43.660 NSThreadAndBlockDemo[1944:12b03] [NSThread sleepForTimeInterval:3];
2013-07-24 16:38:46.660 NSThreadAndBlockDemo[1944:12303] [NSThread sleepForTimeInterval:6];
二、dispatch_group_async的使用
dispatch_group_async可以实现监听一组任务是否完成,完成后得到通知执行其他的操作。这个方法很有用,比如你执行三个下载任务,当三个任务都下载完成后你才通知界面说完成的了。下面是一段例子代码:
[cpp] view plaincopy
(IBAction)touchUpInsideByThreadOne:(id)sender {
dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
dispatch_group_t group = dispatch_group_create();
dispatch_group_async(group, queue, ^{
[NSThread sleepForTimeInterval:6];
NSLog(@"group1 [NSThread sleepForTimeInterval:6];");
});
dispatch_group_async(group, queue, ^{
[NSThread sleepForTimeInterval:3];
NSLog(@"group2 [NSThread sleepForTimeInterval:3];");
});
dispatch_group_async(group, queue, ^{
[NSThread sleepForTimeInterval:1];
NSLog(@"group3 [NSThread sleepForTimeInterval:1];");
});
dispatch_group_notify(group, dispatch_get_main_queue(), ^{
NSLog(@"main thread.");
});
dispatch_release(group);
}
执行结果为: [cpp] view plaincopy 2013-07-24 16:48:23.063 NSThreadAndBlockDemo[2004:12e03] group3 [NSThread sleepForTimeInterval:1];
2013-07-24 16:48:25.063 NSThreadAndBlockDemo[2004:12b03] group2 [NSThread sleepForTimeInterval:3];
2013-07-24 16:48:28.063 NSThreadAndBlockDemo[2004:12303] group1 [NSThread sleepForTimeInterval:6];
2013-07-24 16:48:28.065 NSThreadAndBlockDemo[2004:11303] main thread.
果然,dispatch_group_async只会监听最终的结果完成后,并通知main queue,那如果是我们需要顺序执行的话呢?请看下面的dispatch_barrier_async。
3、dispatch_barrier_async的使用
dispatch_barrier_async是在前面的任务执行结束后它才执行,而且它后面的任务等它执行完成之后才会执行
例子代码如下:
[cpp] view plaincopy
(IBAction)touchUpInsideByThreadOne:(id)sender {
dispatch_queue_t queue = dispatch_queue_create("gcdtest.rongfzh.yc", DISPATCH_QUEUE_CONCURRENT);
dispatch_async(queue, ^{
[NSThread sleepForTimeInterval:3];
NSLog(@"dispatch_async1");
});
dispatch_async(queue, ^{
[NSThread sleepForTimeInterval:1];
NSLog(@"dispatch_async2");
});
dispatch_barrier_async(queue, ^{
NSLog(@"dispatch_barrier_async");
[NSThread sleepForTimeInterval:0.5];
});
dispatch_async(queue, ^{
[NSThread sleepForTimeInterval:1];
NSLog(@"dispatch_async3");
});
}
执行结果为: [cpp] view plaincopy 2013-07-24 17:01:54.580 NSThreadAndBlockDemo[2153:12b03] dispatch_async2
2013-07-24 17:01:56.580 NSThreadAndBlockDemo[2153:12303] dispatch_async1
2013-07-24 17:01:56.580 NSThreadAndBlockDemo[2153:12303] dispatch_barrier_async
2013-07-24 17:01:58.083 NSThreadAndBlockDemo[2153:12303] dispatch_async3
如果使用dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);会发现运行结果为: [cpp] view plaincopy 2013-07-24 17:07:17.577 NSThreadAndBlockDemo[2247:12e03] dispatch_barrier_async
2013-07-24 17:07:18.579 NSThreadAndBlockDemo[2247:15207] dispatch_async3
2013-07-24 17:07:19.578 NSThreadAndBlockDemo[2247:12b03] dispatch_async2
2013-07-24 17:07:20.577 NSThreadAndBlockDemo[2247:12303] dispatch_async1
说明dispatch_barrier_async的顺序执行还是依赖queue的类型啊,必需要queue的类型为dispatch_queue_create创建的,而且attr参数值必需是DISPATCH_QUEUE_CONCURRENT类型,前面两个非dispatch_barrier_async的类型的执行是依赖其本身的执行时间的,如果attr如果是DISPATCH_QUEUE_SERIAL时,那就完全是符合Serial queue的FIFO特征了。
4、dispatch_apply
执行某个代码片段N次。
dispatch_apply(5, globalQ, ^(size_t index) {
// 执行5次
});
5、dispatch_once
dispatch_once这个函数,它可以保证整个应用程序生命周期中某段代码只被执行一次!
[cpp] view plaincopy static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^{
// code to be executed once
});
6、dispatch_after 有时候我们需要等个几秒钟然后做个动画或者给个提示,这时候可以用dispatch_after这个函数: [cpp] view plaincopy double delayInSeconds = 2.0;
dispatch_time_t popTime = dispatch_time(DISPATCH_TIME_NOW, delayInSeconds * NSEC_PER_SEC);
dispatch_after(popTime, dispatch_get_main_queue(), ^(void){
// code to be executed on the main queue after delay
});
7、dispatch_set_target_queue 通过dispatch_set_target_queue函数可以设置一个dispatch queue的优先级,或者指定一个dispatch source相应的事件处理提交到哪个queue上。 [cpp] view plaincopy dispatch_set_target_queue(serialQ, globalQ);
由此可见,GCD的使用非常简单,以我的使用经验来看,以后会逐步淘汰使用NSOperation而改用GCD.