CCTextureCache的多线程加载原理和使用
做引擎的时候,遇到一个texture的异步加载,这里将具体的原理和使用方法贴出来,后面根据浏览器的特性做修改移植。
voidCCTextureCache::addImageAsync(constchar *path, CCObject *target, SEL_CallFuncO selector)
{
CCAssert(path != NULL, "TextureCache: fileimage MUST not be NULL");
CCTexture2D *texture = NULL;
// optimization
std::string pathKey = path;
CCFileUtils::ccRemoveHDSuffixFromFile(pathKey);
pathKey = CCFileUtils::fullPathFromRelativePath(pathKey.c_str());
texture = m_pTextures->objectForKey(pathKey); //根据pathkey查看是否纹理已经加载过,如果已经有了,则不重复加载
std::string fullpath = pathKey;
if (texture != NULL)
{
if (target && selector)
{
(target->*selector)(texture);
}
return;
}
if (target)
{
target->retain();
}
// lazy init
static bool firstRun = true;
if (firstRun)
{
s_pAsyncStructQueue = new queue<AsyncStruct*>();
s_pImageQueue = new queue<ImageInfo*>();
pthread_mutex_init(&s_asyncStructQueueMutex, NULL);
sem_init(&s_sem, 0, 0);
pthread_mutex_init(&s_ImageInfoMutex, NULL);
pthread_create(&s_loadingThread, NULL, loadImage, NULL); //创建新的线程,用于后台加载图片
//创建调度队列,用来根据已加载的图片进行纹理转换
CCScheduler::sharedScheduler()->scheduleSelector(schedule_selector(CCTextureCache::addImageAsyncCallBack), this, 0, false);
need_quit = false;
firstRun = false;
}
// generate async struct
AsyncStruct *data = new AsyncStruct();
data->filename = fullpath.c_str();
data->target = target;
data->selector = selector;
// add async struct into queue
pthread_mutex_lock(&s_asyncStructQueueMutex);
s_pAsyncStructQueue->push(data); //将需要加载的图片放入队列中
pthread_mutex_unlock(&s_asyncStructQueueMutex);
sem_post(&s_sem);
}
从上述代码分析可以看出其过程:
1.创建线程,用于后台加载
2. 将对于需要加载的图片放入队列中
3. callback函数设定,用于将加载完成的图片转为纹理,等待使用其调用是由CCTimer::update调用的。
4. addImageAsyncCallBack函数在处理完纹理转换,还会调用addImageAsync传入的SEL_CallFuncO selector,实现用户加载图片纹理之后的具体处理。
使用例子:
CCTextureCache::sharedTextureCache()->addImageAsync("Images/blocks.png", this, callfuncO_selector(TextureCacheTest::loadingCallBack));
loadingCallBack函数就是使用异步加载完之后的用户可以自处理结果,比如初始化一个sprite,用这个纹理贴出来。