您还未登录! 登录 | 注册 | 帮助  

您的位置: 首页 > 软件测试技术 > 测试用例 > 正文

关于GCD的几个用例总结

发表于:2017-01-09 作者:网络转载 来源:

  GCD作为日常开发中的使用已经非常普遍,基于C的API为应用在多核硬件上高效运行提供了有力支持。本文分场景写了几个测试Demo,方便大家理解与应用。
  1,dispatch_get_global_queue与dispatch_get_main_queue交互
  很多应用场景需要后台读写大量数据,通过dispatch_get_global_queue函数可以获取全局队列来并发执行后台任务,并再结束后更新UI,保证应用的流程,避免主线程阻塞。代码如下:
- (void)test1
{
dispatch_queue_t queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
dispatch_group_t group = dispatch_group_create();
for (int i = 0; i < 10; i++) {
dispatch_group_async(group, queue, ^{
NSLog(@"___%i %@", i, [NSThread currentThread]);
});
}
dispatch_group_notify(group, queue, ^{
NSLog(@"currentThread1 currentThread]);
NSLog(@"执行完毕");
dispatch_async(dispatch_get_main_queue(), ^{
NSLog(@"currentThread2
currentThread]);
NSLog(@"main_queue 执行完毕");
});
});
}
  执行结果:
  test1:
  2016-03-10 17:21:37.293 GCDSample[3892:2238505] ___0 {number = 2, name = (null)}
  2016-03-10 17:21:37.293 GCDSample[3892:2238506] ___2 {number = 4, name = (null)}
  2016-03-10 17:21:37.293 GCDSample[3892:2238504] ___3 {number = 3, name = (null)}
  2016-03-10 17:21:37.293 GCDSample[3892:2238507] ___1 {number = 5, name = (null)}
  2016-03-10 17:21:37.294 GCDSample[3892:2238505] ___4 {number = 2, name = (null)}
  2016-03-10 17:21:37.295 GCDSample[3892:2238506] ___5 {number = 4, name = (null)}
  2016-03-10 17:21:37.295 GCDSample[3892:2238504] ___6 {number = 3, name = (null)}
  2016-03-10 17:21:37.295 GCDSample[3892:2238507] ___7 {number = 5, name = (null)}
  2016-03-10 17:21:37.295 GCDSample[3892:2238505] ___8 {number = 2, name = (null)}
  2016-03-10 17:21:37.298 GCDSample[3892:2238506] ___9 {number = 4, name = (null)}
  2016-03-10 17:21:37.299 GCDSample[3892:2238506] currentThread1 {number = 4, name = (null)}
  2016-03-10 17:21:37.299 GCDSample[3892:2238506] 执行完毕
  2016-03-10 17:21:37.302 GCDSample[3892:2238481] currentThread2 {number = 1, name = main}
  2016-03-10 17:21:37.303 GCDSample[3892:2238481] main_queue 执行完毕
  2,异步线程中串行执行任务
  test1中dispatch_group_async以并发的方式开启异步线程,不能保证执行顺序,如果想在并发线程中串行执行任务该如何做呢?只需要创建一个串行队列,加入group任务即可。
- (void)test1
{
NSLog(@"code begin");
NSLog(@"currentThread0
currentThread]);
dispatch_group_t group = dispatch_group_create();
dispatch_queue_t serialQueue = dispatch_queue_create(NULL, DISPATCH_QUEUE_SERIAL);
for (int i = 0; i < 10; i++) {
dispatch_group_async(group, serialQueue, ^{
[NSThread sleepForTimeInterval:.1f];
NSLog(@"___%i %@", i, [NSThread currentThread]);
});
}
dispatch_group_notify(group, serialQueue, ^{
NSLog(@"currentThread1
currentThread]);
NSLog(@"执行完毕");
});
NSLog(@"code end");
}
  执行结果:
  test2:
  2016-03-10 17:07:10.366 GCDSample[3879:2235420] code begin
  2016-03-10 17:07:10.367 GCDSample[3879:2235420] currentThread0 {number = 1, name = main}
  2016-03-10 17:07:10.367 GCDSample[3879:2235420] code end
  2016-03-10 17:07:10.472 GCDSample[3879:2235438] ___0 {number = 2, name = (null)}
  2016-03-10 17:07:10.577 GCDSample[3879:2235438] ___1 {number = 2, name = (null)}
  2016-03-10 17:07:10.683 GCDSample[3879:2235438] ___2 {number = 2, name = (null)}
  2016-03-10 17:07:10.784 GCDSample[3879:2235438] ___3 {number = 2, name = (null)}
  2016-03-10 17:07:10.890 GCDSample[3879:2235438] ___4 {number = 2, name = (null)}
  2016-03-10 17:07:10.995 GCDSample[3879:2235438] ___5 {number = 2, name = (null)}
  2016-03-10 17:07:11.101 GCDSample[3879:2235438] ___6 {number = 2, name = (null)}
  2016-03-10 17:07:11.206 GCDSample[3879:2235438] ___7 {number = 2, name = (null)}
  2016-03-10 17:07:11.309 GCDSample[3879:2235438] ___8 {number = 2, name = (null)}
  2016-03-10 17:07:11.411 GCDSample[3879:2235438] ___9 {number = 2, name = (null)}
  2016-03-10 17:07:11.411 GCDSample[3879:2235438] currentThread1 {number = 2, name = (null)}
  2016-03-10 17:07:11.411 GCDSample[3879:2235438] 执行完毕
  3,dispatch_barrier_async,分割多任务线程
  如果有10个并发任务,想要分成两组,比如指定前五个任务执行完之后,优先执行第六个任务,再执行剩下的操作,可以通过dispatch_barrier_async来分割开,示例:
- (void)test3
{
dispatch_queue_t concurrentQueue = ({
dispatch_queue_t queue = dispatch_queue_create("concurrentQueue", DISPATCH_QUEUE_CONCURRENT);
queue;
});
for (int i = 0; i < 10; i++) {
if (i != 5) {
dispatch_async(concurrentQueue, ^{
[NSThread sleepForTimeInterval:arc4random_uniform(3)];
NSLog(@"dispatch_async %i", i);
});
}else{
dispatch_barrier_async(concurrentQueue, ^{
NSLog(@"dispatch_barrier_async");
});
}
}
}
  执行结果:
  test3
  2016-03-10 17:47:32.397 GCDSample[3949:2246479] dispatch_async 1
  2016-03-10 17:47:32.400 GCDSample[3949:2246481] dispatch_async 3
  2016-03-10 17:47:33.403 GCDSample[3949:2246480] dispatch_async 2
  2016-03-10 17:47:34.400 GCDSample[3949:2246477] dispatch_async 0
  2016-03-10 17:47:34.405 GCDSample[3949:2246479] dispatch_async 4
  2016-03-10 17:47:34.406 GCDSample[3949:2246477] dispatch_barrier_async
  2016-03-10 17:47:34.406 GCDSample[3949:2246479] dispatch_async 6
  2016-03-10 17:47:35.411 GCDSample[3949:2246479] dispatch_async 7
  2016-03-10 17:47:35.412 GCDSample[3949:2246480] dispatch_async 9
  2016-03-10 17:47:35.411 GCDSample[3949:2246477] dispatch_async 8
  4,串行队列死锁
- (void)test4
{
NSLog(@"1");
dispatch_sync(dispatch_get_main_queue(), ^{
NSLog(@"2");
});
NSLog(@"3");
}
- (void)test4_1 {
NSLog(@"1");
dispatch_async(dispatch_get_global_queue(0, 0), ^{
NSLog(@"2
currentThread]);
dispatch_sync(dispatch_get_main_queue(), ^{
NSLog(@"3
currentThread]);
});
NSLog(@"4
currentThread]);
});
NSLog(@"5");
}
  方法test4会造成死锁,因为main queue需要等待dispatch_sync函数中block返回才能继续执行,而通过dispatch_sync放入main queue的block按照FIFO的原则(先入先出)现在得不到执行,就造成了相互等待的局面,产生死锁。将同步的串行队列放到另外一个异步线程就能够解决(如方法test4_1所示)。所以在使用dispatch_sync的时候需要很谨慎,需要先执行[NSThread isMainThread]判断下当前任务是否在mian queue中调用,比如有一段代码在后台执行,而它需要从界面控制层获取一个值。那么你可以使用dispatch_sync简单办到。执行结果:
  test4
  2016-03-10 17:23:30.020 GCDSample[3896:2239119] 1
  test4_1
  2016-03-11 17:30:11.900 GCDSample[1197:730552] 1
  2016-03-11 17:30:11.901 GCDSample[1197:730552] 5
  2016-03-11 17:30:11.903 GCDSample[1197:730564] 2 {number = 2, name = (null)}
  2016-03-11 17:30:11.933 GCDSample[1197:730552] 3 {number = 1, name = main}
  2016-03-11 17:30:11.933 GCDSample[1197:730564] 4 {number = 2, name = (null)}

  5,dispatch_semaphore_t 控制并发线程数
  有时在执行多任务时需要避免抢占资源以及性能过多消耗的情况,需要在特定时间内控制同时执行的任务数量,在NSOperationQueue可以通过maxConcurrentOperationCount来控制,在GCD中可以指定semaphore来控制了。先介绍3个函数,dispatch_semaphore_create创建一个semaphore;dispatch_semaphore_wait等待信号,当信号总量少于0的时候就会一直等待,否则就可以正常的执行,并让信号总量-1;dispatch_semaphore_signal是发送一个信号,自然会让信号总量加1。看一个小例子:
- (void)test7
{
_semaphore = dispatch_semaphore_create(2);
[self task7_1];
[self task7_2];
[self task7_3];
[self task7_4];
}
- (void)task7_1
{
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
NSLog(@"7_1_begin");
sleep(4);
NSLog(@"7_1_end");
dispatch_semaphore_signal(_semaphore);
});
}
- (void)task7_2
{
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
NSLog(@"7_2_begin");
sleep(4);
NSLog(@"7_2_end");
dispatch_semaphore_signal(_semaphore);
});
}
- (void)task7_3
{
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
NSLog(@"7_3_begin");
sleep(4);
NSLog(@"7_3_end");
dispatch_semaphore_signal(_semaphore);
});
}
- (void)task7_4
{
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
NSLog(@"7_4_begin");
sleep(4);
NSLog(@"7_4_end");
dispatch_semaphore_signal(_semaphore);
});
}
  执行结果:
  semaphore 为 1
  2016-03-11 16:33:11.957 GCDSample[22394:2943718] 7_1_begin
  2016-03-11 16:33:15.959 GCDSample[22394:2943718] 7_1_end
  2016-03-11 16:33:15.960 GCDSample[22394:2943717] 7_2_begin
  2016-03-11 16:33:19.964 GCDSample[22394:2943717] 7_2_end
  2016-03-11 16:33:19.965 GCDSample[22394:2943721] 7_3_begin
  2016-03-11 16:33:23.970 GCDSample[22394:2943721] 7_3_end
  2016-03-11 16:33:23.970 GCDSample[22394:2943727] 7_4_begin
  2016-03-11 16:33:27.975 GCDSample[22394:2943727] 7_4_end
  ------- ------- ------- ------- ------- ------- ------- ----
  semaphore 为 2
  2016-03-11 16:33:47.396 GCDSample[22406:2944975] 7_2_begin
  2016-03-11 16:33:47.396 GCDSample[22406:2944974] 7_1_begin
  2016-03-11 16:33:51.398 GCDSample[22406:2944975] 7_2_end
  2016-03-11 16:33:51.398 GCDSample[22406:2944974] 7_1_end
  2016-03-11 16:33:51.398 GCDSample[22406:2944976] 7_3_begin
  2016-03-11 16:33:51.398 GCDSample[22406:2944983] 7_4_begin
  2016-03-11 16:33:55.401 GCDSample[22406:2944976] 7_3_end
  2016-03-11 16:33:55.401 GCDSample[22406:2944983] 7_4_end
  可以看到,并发执行任务的数量取决于_semaphore = dispatch_semaphore_create(2);传入的数字,当为1时,效果如同执行串行队列。
  6,dispatch_set_target_queue 设置队列优先级
  如果有串行队列A和并行队列B,队列A中加入任务1,队列B中加入任务2、任务3,如果确保1、2、3顺序执行呢?可以通过dispatch_set_target_queue设置队列的优先级,将队列AB指派到队列C上,任务123将会在串行队列C中顺序执行。代码如下:
- (void)test8
{
dispatch_queue_t serialQueue = dispatch_queue_create("com.starming.gcddemo.serialqueue", DISPATCH_QUEUE_SERIAL);
dispatch_queue_t firstQueue = dispatch_queue_create("com.starming.gcddemo.firstqueue", DISPATCH_QUEUE_SERIAL);
dispatch_queue_t secondQueue = dispatch_queue_create("com.starming.gcddemo.secondqueue", DISPATCH_QUEUE_CONCURRENT);
dispatch_set_target_queue(firstQueue, serialQueue);
dispatch_set_target_queue(secondQueue, serialQueue);
dispatch_async(firstQueue, ^{
NSLog(@"1
currentThread]);
[NSThread sleepForTimeInterval:3.f];
});
dispatch_async(secondQueue, ^{
NSLog(@"2
currentThread]);
[NSThread sleepForTimeInterval:2.f];
});
dispatch_async(secondQueue, ^{
NSLog(@"3
currentThread]);
[NSThread sleepForTimeInterval:1.f];
});
}
  执行结果:
  2016-03-11 17:31:41.515 GCDSample[1202:730942] 1 {number = 2, name = (null)}
  2016-03-11 17:31:44.518 GCDSample[1202:730942] 2 {number = 2, name = (null)}
  2016-03-11 17:31:46.520 GCDSample[1202:730942] 3 {number = 2, name = (null)}
  未完待续,Have fun!