cocos2d-x游戏引擎核心(3.x)----启动渲染流程

(1) 首先,这里以win32平台下为例子.win32下游戏的启动都是从win32目录下main文件开始的,即是游戏的入口函数,如下:

#include "main.h"
#include "AppDelegate.h"
#include "cocos2d.h"

USING_NS_CC;

int APIENTRY _tWinMain(HINSTANCE hInstance,
                       HINSTANCE hPrevInstance,
                       LPTSTR    lpCmdLine,
                       int       nCmdShow)
{
    UNREFERENCED_PARAMETER(hPrevInstance);
    UNREFERENCED_PARAMETER(lpCmdLine);

    // create the application instance
    AppDelegate app;
　　 // 启动游戏
    return Application::getInstance()->run();
}

(1-1)这里可以看出,在入口函数中,首先创建了一个AppDelegate对象，AppDelegate继承 自CCApplication，在创建APPDelegate对象的时候就会隐式调用CCApplication构造函数，在这个构造函数里边会将AppDelegate的this指针传递给全局共享对象sm_pSharedApplication,如下:

Application::Application()
//初始化win32应用程序对象 
: _instance(nullptr)
, _accelTable(nullptr)
{
    _instance    = GetModuleHandle(nullptr);
　　// 用于控制帧数的计数值
    _animationInterval.QuadPart = 0;
    CC_ASSERT(! sm_pSharedApplication);
    // 全局共享对象
    sm_pSharedApplication = this;
}

(1-2) 接下来调用Application::getInstance()->run();启动游戏,如下:

int Application::run()
{
    PVRFrameEnableControlWindow(false);

    // Main message loop:
    LARGE_INTEGER nFreq;
    LARGE_INTEGER nLast;
    LARGE_INTEGER nNow;

    QueryPerformanceFrequency(&nFreq);
    QueryPerformanceCounter(&nLast);

    // Initialize instance and cocos2d.
    // 执行AppDeletegate重载的applicationDidFinishLaunching函数
    if (!applicationDidFinishLaunching())
    {
        return 0;
    }

    auto director = Director::getInstance();
    auto glview = director->getOpenGLView();

    // Retain glview to avoid glview being released in the while loop
    glview->retain();

    while(!glview->windowShouldClose())
    {
        QueryPerformanceCounter(&nNow);
        if (nNow.QuadPart - nLast.QuadPart > _animationInterval.QuadPart)
        {
            nLast.QuadPart = nNow.QuadPart;
            // 主循环,每帧调用
            director->mainLoop();
            glview->pollEvents();
        }
        else
        {
            Sleep(0);
        }
    }

    // Director should still do a cleanup if the window was closed manually.
    if (glview->isOpenGLReady())
    {
        // 结束,执行清理工作
        director->end();
        director->mainLoop();
        director = nullptr;
    }
    glview->release();
    return true;
}

(1-2-1) 我们进入到AppDelegate::applicationDidFinishLaunching(),看它究竟做了什么,我们以/cocos2d-x-3.2/templates/cpp-template-default/Classes/AppDelegate.cpp为例:

bool AppDelegate::applicationDidFinishLaunching() {
    // initialize director
    auto director = Director::getInstance();
    auto glview = director->getOpenGLView();
    if(!glview) {
        // 创建glview对象, 这里采用默认的分辨率先创建出游戏窗口
        glview = GLView::create("My Game");
        // 这里设置了和OpenGL相关的一些信息
        director->setOpenGLView(glview);
    }

    // turn on display FPS
    director->setDisplayStats(true);

    // set FPS. the default value is 1.0/60 if you don't call this
    director->setAnimationInterval(1.0 / 60);

    // create a scene. it's an autorelease object
    // 创建场景
    auto scene = HelloWorld::createScene();

    // run 运行场景
    director->runWithScene(scene);

    return true;
}

(1-2-1-1) 可以看到applicationDidFinishLaunching函数里面设置了glview对象之后,就开始运行场景,可以进入GLView::create中看其究竟是如何创建GLView对象,同样,我们是win32下面看的, 所以找到cocos2d-x-3.2/cocos/platform/desktop/CCGLView.cpp文件:

GLView* GLView::create(const std::string& viewName)
{
    auto ret = new GLView;
    if(ret && ret->initWithRect(viewName, Rect(0, 0, 960, 640), 1)) {
        ret->autorelease();
        return ret;
    }

    return nullptr;
}

从代码可以看到只是简单的new一个GLView对象,我们进入/cocos2d-x-3.2/cocos/platform/desktop/CCGLView.h看一下它究竟是个什么东西:

/****************************************************************************
Copyright (c) 2010-2012 cocos2d-x.org
Copyright (c) 2013-2014 Chukong Technologies Inc.

http://www.cocos2d-x.org

*/

#ifndef __CC_EGLVIEW_DESKTOP_H__
#define __CC_EGLVIEW_DESKTOP_H__

#include "base/CCRef.h"
#include "platform/CCCommon.h"
#include "platform/CCGLViewProtocol.h"
#include "glfw3.h"

NS_CC_BEGIN

class CC_DLL GLView : public GLViewProtocol, public Ref
{
public:
    static GLView* create(const std::string& viewName);
    static GLView* createWithRect(const std::string& viewName, Rect size, float frameZoomFactor = 1.0f);
    static GLView* createWithFullScreen(const std::string& viewName);
    static GLView* createWithFullScreen(const std::string& viewName, const GLFWvidmode &videoMode, GLFWmonitor *monitor);

    /*
     *frameZoomFactor for frame. This method is for debugging big resolution (e.g.new ipad) app on desktop.
     */

    //void resize(int width, int height);

    float getFrameZoomFactor();
    //void centerWindow();

    virtual void setViewPortInPoints(float x , float y , float w , float h);
    virtual void setScissorInPoints(float x , float y , float w , float h);


    bool windowShouldClose();
    void pollEvents();
    GLFWwindow* getWindow() const { return _mainWindow; }

    /* override functions */
    virtual bool isOpenGLReady() override;
    // 删除窗口，做窗口清理工作
    virtual void end() override;
    // 交换buffer
    virtual void swapBuffers() override;
    // 设置窗口大小
    virtual void setFrameSize(float width, float height) override;
    // 设置输入法状态
    virtual void setIMEKeyboardState(bool bOpen) override;

    /*
     * Set zoom factor for frame. This method is for debugging big resolution (e.g.new ipad) app on desktop.
     */
    void setFrameZoomFactor(float zoomFactor);

    /** Retina support is disabled by default
     *  @note This method is only available on Mac.
     */
    void enableRetina(bool enabled);
    /** Check whether retina display is enabled. */
    bool isRetinaEnabled() const { return _isRetinaEnabled; };
    
    /** Get retina factor */
    int getRetinaFactor() const { return _retinaFactor; }

protected:
    GLView();
    virtual ~GLView();

    bool initWithRect(const std::string& viewName, Rect rect, float frameZoomFactor);
    bool initWithFullScreen(const std::string& viewName);
    bool initWithFullscreen(const std::string& viewname, const GLFWvidmode &videoMode, GLFWmonitor *monitor);

    bool initGlew();

    void updateFrameSize();

    // GLFW callbacks
    void onGLFWError(int errorID, const char* errorDesc);
    void onGLFWMouseCallBack(GLFWwindow* window, int button, int action, int modify);
    void onGLFWMouseMoveCallBack(GLFWwindow* window, double x, double y);
    void onGLFWMouseScrollCallback(GLFWwindow* window, double x, double y);
    void onGLFWKeyCallback(GLFWwindow* window, int key, int scancode, int action, int mods);
    void onGLFWCharCallback(GLFWwindow* window, unsigned int character);
    void onGLFWWindowPosCallback(GLFWwindow* windows, int x, int y);
    void onGLFWframebuffersize(GLFWwindow* window, int w, int h);
    void onGLFWWindowSizeFunCallback(GLFWwindow *window, int width, int height);

    bool _captured;
    bool _supportTouch;
    bool _isInRetinaMonitor;
    bool _isRetinaEnabled;
    int  _retinaFactor;  // Should be 1 or 2

    float _frameZoomFactor;

    GLFWwindow* _mainWindow;
    GLFWmonitor* _monitor;

    float _mouseX;
    float _mouseY;

    friend class GLFWEventHandler;

private:
    CC_DISALLOW_COPY_AND_ASSIGN(GLView);
};

NS_CC_END   // end of namespace   cocos2d

#endif  // end of __CC_EGLVIEW_DESKTOP_H__

GLView继承自GLViewProtocol,我们也进入看一下:

/****************************************************************************
Copyright (c) 2010-2012 cocos2d-x.org
Copyright (c) 2013-2014 Chukong Technologies Inc.

http://www.cocos2d-x.org

*******************************************************/

#ifndef __CCGLVIEWPROTOCOL_H__
#define __CCGLVIEWPROTOCOL_H__

#include "base/ccTypes.h"
#include "base/CCEventTouch.h"

#include <vector>
// 5种屏幕适配策略
enum class ResolutionPolicy
{
    EXACT_FIT,
    NO_BORDER,
    SHOW_ALL,
    FIXED_HEIGHT,
    FIXED_WIDTH,

    UNKNOWN,
};

NS_CC_BEGIN

class CC_DLL GLViewProtocol
{
public:
    /**
     * @js ctor
     */
    GLViewProtocol();
    /**
     * @js NA
     * @lua NA
     */
    virtual ~GLViewProtocol();

    /** Force destroying EGL view, subclass must implement this method. */
    virtual void end() = 0;

    /** Get whether opengl render system is ready, subclass must implement this method. */
    virtual bool isOpenGLReady() = 0;

    /** Exchanges the front and back buffers, subclass must implement this method. */
    virtual void swapBuffers() = 0;

    /** Open or close IME keyboard , subclass must implement this method. */
    virtual void setIMEKeyboardState(bool open) = 0;

    /**
     * Polls input events. Subclass must implement methods if platform
     * does not provide event callbacks.
     */
    virtual void pollInputEvents();

    /**
     * Get the frame size of EGL view.
     * In general, it returns the screen size since the EGL view is a fullscreen view.
     */
    virtual const Size& getFrameSize() const;

    /**
     * Set the frame size of EGL view.
     */
    virtual void setFrameSize(float width, float height);

    // 获取可见区域的原点和大小
    virtual Size getVisibleSize() const;
    virtual Vec2 getVisibleOrigin() const;
    virtual Rect getVisibleRect() const;

    //设置设计的size，当需要适配多种设备时，可以用这个函数定义逻辑坐标，cocos2dx会自动将逻辑坐标转化成实际坐标,这样一样的代码可以适配各种设备分辨率
    virtual void setDesignResolutionSize(float width, float height, ResolutionPolicy resolutionPolicy);

    /** Get design resolution size.
     *  Default resolution size is the same as 'getFrameSize'.
     */
    virtual const Size&  getDesignResolutionSize() const;

    /**
     * Set opengl view port rectangle with points.
     */
    virtual void setViewPortInPoints(float x , float y , float w , float h);

    /**
     * Set Scissor rectangle with points.
     */
    virtual void setScissorInPoints(float x , float y , float w , float h);

    /**
     * Get whether GL_SCISSOR_TEST is enable
     */
    virtual bool isScissorEnabled();

    /**
     * Get the current scissor rectangle
     */
    virtual Rect getScissorRect() const;

    virtual void setViewName(const std::string& viewname);
    const std::string& getViewName() const;

    /** Touch events are handled by default; if you want to customize your handlers, please override these functions: */
    // 触摸处理函数，可以重载
    virtual void handleTouchesBegin(int num, intptr_t ids[], float xs[], float ys[]);
    virtual void handleTouchesMove(int num, intptr_t ids[], float xs[], float ys[]);
    virtual void handleTouchesEnd(int num, intptr_t ids[], float xs[], float ys[]);
    virtual void handleTouchesCancel(int num, intptr_t ids[], float xs[], float ys[]);

    /**
     * Get the opengl view port rectangle.
     */
    const Rect& getViewPortRect() const;

    /**
     * Get scale factor of the horizontal direction.
     */
    float getScaleX() const;

    /**
     * Get scale factor of the vertical direction.
     */
    float getScaleY() const;

    /** returns the current Resolution policy */
    ResolutionPolicy getResolutionPolicy() const { return _resolutionPolicy; }

protected:
    void updateDesignResolutionSize();
    
    void handleTouchesOfEndOrCancel(EventTouch::EventCode eventCode, int num, intptr_t ids[], float xs[], float ys[]);

    // real screen size
    Size _screenSize;
    // resolution size, it is the size appropriate for the app resources.
    Size _designResolutionSize;
    // the view port size
    Rect _viewPortRect;
    // the view name
    std::string _viewName;

    float _scaleX;
    float _scaleY;
    ResolutionPolicy _resolutionPolicy;
};

// end of platform group
/// @}

NS_CC_END

#endif /* __CCGLVIEWPROTOCOL_H__ */

以看到CCEGLView和GLViewProtocol是显示窗口，负责窗口级别的功能管理和实现, 包括：坐标和缩放管理, 画图工具,按键事件;

(1-2-1-2) 创建glview对象之后,导演类Director就把glview设置进游戏,其中包括很多配置信息, 如设置屏幕大小适配相关的函数getDesignResolutionSize, 如下:

void Director::setOpenGLView(GLView *openGLView)
{
    CCASSERT(openGLView, "opengl view should not be null");

    if (_openGLView != openGLView)
    {
        // Configuration. Gather GPU info
        Configuration *conf = Configuration::getInstance();
        conf->gatherGPUInfo();
        CCLOG("%s
",conf->getInfo().c_str());

        if(_openGLView)
            _openGLView->release();
        _openGLView = openGLView;
        _openGLView->retain();

        // set size 设置屏幕大小适配相关的函数
        _winSizeInPoints = _openGLView->getDesignResolutionSize();

        createStatsLabel();

        if (_openGLView)
        {
            setGLDefaultValues();
        }
       // 完成初始化
        _renderer->initGLView();

        CHECK_GL_ERROR_DEBUG();

        if (_eventDispatcher)
        {
            _eventDispatcher->setEnabled(true);
        }
    }
}

(1-2-1-2-1) 我们进入initGLView看看它都做了什么初始化工作,找到/cocos2d-x-3.2/cocos/renderer/CCRenderer.cpp:

void Renderer::initGLView()
{
#if CC_ENABLE_CACHE_TEXTURE_DATA
    _cacheTextureListener = EventListenerCustom::create(EVENT_RENDERER_RECREATED, [this](EventCustom* event){
        /** listen the event that renderer was recreated on Android/WP8 */
        this->setupBuffer();
    });
    
    Director::getInstance()->getEventDispatcher()->addEventListenerWithFixedPriority(_cacheTextureListener, -1);
#endif
    // 填充索引缓冲
    setupIndices();
    
    setupBuffer();
    
    _glViewAssigned = true;
}

(1-2-1-2-1-1) 进入setupIndices如下:

void Renderer::setupIndices()
{
    for( int i=0; i < VBO_SIZE; i++)
    {
　　　　// 计算索引缓冲值
        _indices[i*6+0] = (GLushort) (i*4+0);
        _indices[i*6+1] = (GLushort) (i*4+1);
        _indices[i*6+2] = (GLushort) (i*4+2);
        _indices[i*6+3] = (GLushort) (i*4+3);
        _indices[i*6+4] = (GLushort) (i*4+2);
        _indices[i*6+5] = (GLushort) (i*4+1);
    }
}

(1-2-1-2-1-2) 进入setupBuffer如下:

void Renderer::setupBuffer()
{
    // 如果使用VAO
    if(Configuration::getInstance()->supportsShareableVAO())
    {
        // 初始化VAO和VBO
        setupVBOAndVAO();
    }
    else
    {
        // 初始化VBO
        setupVBO();
    }
}

(1-2-1-2-1-2-1) 进入setupVBOAndVAO和setupVBO, 开始调用OpenGL API进行顶点数据指定,具体意义参见基于Cocos2d-x学习OpenGL ES 2.0系列——编写自己的shader（2）:

void Renderer::setupVBOAndVAO()
{
    glGenVertexArrays(1, &_quadVAO);
    GL::bindVAO(_quadVAO);

    glGenBuffers(2, &_buffersVBO[0]);

    glBindBuffer(GL_ARRAY_BUFFER, _buffersVBO[0]);
    glBufferData(GL_ARRAY_BUFFER, sizeof(_quads[0]) * VBO_SIZE, _quads, GL_DYNAMIC_DRAW);

    // vertices
    glEnableVertexAttribArray(GLProgram::VERTEX_ATTRIB_POSITION);
    glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_POSITION, 3, GL_FLOAT, GL_FALSE, sizeof(V3F_C4B_T2F), (GLvoid*) offsetof( V3F_C4B_T2F, vertices));

    // colors
    glEnableVertexAttribArray(GLProgram::VERTEX_ATTRIB_COLOR);
    glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_COLOR, 4, GL_UNSIGNED_BYTE, GL_TRUE, sizeof(V3F_C4B_T2F), (GLvoid*) offsetof( V3F_C4B_T2F, colors));

    // tex coords
    glEnableVertexAttribArray(GLProgram::VERTEX_ATTRIB_TEX_COORD);
    glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_TEX_COORD, 2, GL_FLOAT, GL_FALSE, sizeof(V3F_C4B_T2F), (GLvoid*) offsetof( V3F_C4B_T2F, texCoords));

    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, _buffersVBO[1]);
    glBufferData(GL_ELEMENT_ARRAY_BUFFER, sizeof(_indices[0]) * VBO_SIZE * 6, _indices, GL_STATIC_DRAW);

    // Must unbind the VAO before changing the element buffer.
    GL::bindVAO(0);
    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
    glBindBuffer(GL_ARRAY_BUFFER, 0);

    CHECK_GL_ERROR_DEBUG();
}

void Renderer::setupVBO()
{
    glGenBuffers(2, &_buffersVBO[0]);

    mapBuffers();
}

(1-2-2) 在applicationDidFinishLaunching里面创建场景之后,就调用director->mainLoop();开始游戏主循环了.我们进入mainLoop看它做了什么, win32下我们找到cocos2d-x-3.2/cocos/base/CCDirector.cpp:

void DisplayLinkDirector::mainLoop()
{
    if (_purgeDirectorInNextLoop)
    {
        _purgeDirectorInNextLoop = false;
        // 主循环结束,清除工作
        purgeDirector();
    }
    else if (! _invalid)
    {
　　　　 // 渲染场景
        drawScene();
     
        // release the objects
        // 释放对象:内存池里之前通过autorelease加入的对象引用计数减 1.
        PoolManager::getInstance()->getCurrentPool()->clear();
    }
}

mainLoop主要完成三个动作:

　　1 判断是否需要释放 CCDirector，如果需要，则删除 CCDirector 占用的资源。通常，游戏结束时才会执行这个步骤。　　

　　2 调用 drawScene()方法，绘制当前场景并进行其他必要的处理。

　　3 弹出自动回收池，使得这一帧被放入自动回收池的对象全部释放。

(1-2-2-1) 由此可见，mainLoop()把内存管理以外的操作都交给了 drawScene()方法，因此关键的步骤都在 drawScene()方法之中。下面是 drawScene()方法的实现：

// Draw the Scene
void Director::drawScene()
{
    // calculate "global" dt
    // 计算全局帧间时间差 dt
    calculateDeltaTime();
    
    // skip one flame when _deltaTime equal to zero.
    if(_deltaTime < FLT_EPSILON)
    {
        return;
    }

    if (_openGLView)
    {
        _openGLView->pollInputEvents();
    }

    //tick before glClear: issue #533
    if (! _paused)
    {
        // 启动定时器
        _scheduler->update(_deltaTime);
        _eventDispatcher->dispatchEvent(_eventAfterUpdate);
    }

    glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

    /* to avoid flickr, nextScene MUST be here: after tick and before draw.
     XXX: Which bug is this one. It seems that it can't be reproduced with v0.9 */
    if (_nextScene)
    {
        // 如果有,设置下一个场景
        setNextScene();
    }

    // 保存原来的模型视图(ModelView)矩阵
    pushMatrix(MATRIX_STACK_TYPE::MATRIX_STACK_MODELVIEW);

    // draw the scene
    if (_runningScene)
    {
　　　　 // 开始绘制场景
        _runningScene->visit(_renderer, Mat4::IDENTITY, false);
        // 事件分发
        _eventDispatcher->dispatchEvent(_eventAfterVisit);
    }

    // draw the notifications node
    if (_notificationNode)
    {
        // 处理通知节点
        _notificationNode->visit(_renderer, Mat4::IDENTITY, false);
    }

    if (_displayStats)
    {
        showStats();
    }
    // 开始渲染场景
    _renderer->render();
    _eventDispatcher->dispatchEvent(_eventAfterDraw);

    popMatrix(MATRIX_STACK_TYPE::MATRIX_STACK_MODELVIEW);

    _totalFrames++;

    // swap buffers 交换缓冲区
    if (_openGLView)
    {
        _openGLView->swapBuffers();
    }

    if (_displayStats)
    {
        calculateMPF();
    }
}

可以发现drawScene主要用于处理 OpenGL 和一些细节，如计算 FPS、帧间时间差等,这里我们主要进行了以下 3 个操作。
　　1 调用了定时调度器的 update 方法，引发定时器事件。
　　2 如果场景需要被切换，则调用 setNextScene 方法，在显示场景前切换场景。
　　3 调用当前场景的 visit 方法，将当前场景加入渲染队列,并通过render统一渲染。

(1-2-2-1-1) 我们进入到visit方法里面,看它怎样把每一个节点添加到渲染队列, 这里我们找到/cocos2d-x-3.2/cocos/2d/CCNode.cpp:

void Node::visit(Renderer* renderer, const Mat4 &parentTransform, uint32_t parentFlags)
{
    // quick return if not visible. children won't be drawn.
    if (!_visible)
    {
        return;
    }

    // 设置_modelViewTransform矩阵
    uint32_t flags = processParentFlags(parentTransform, parentFlags);

    // IMPORTANT:
    // To ease the migration to v3.0, we still support the Mat4 stack,
    // but it is deprecated and your code should not rely on it
    Director* director = Director::getInstance();
    director->pushMatrix(MATRIX_STACK_TYPE::MATRIX_STACK_MODELVIEW);
    director->loadMatrix(MATRIX_STACK_TYPE::MATRIX_STACK_MODELVIEW, _modelViewTransform);

    int i = 0;

    if(!_children.empty())
    {
        sortAllChildren();
        // draw children zOrder < 0
        for( ; i < _children.size(); i++ )
        {
            auto node = _children.at(i);

            if ( node && node->_localZOrder < 0 )
                node->visit(renderer, _modelViewTransform, flags);
            else
                break;
        }
        // self draw
        this->draw(renderer, _modelViewTransform, flags);

        for(auto it=_children.cbegin()+i; it != _children.cend(); ++it)
            (*it)->visit(renderer, _modelViewTransform, flags);
    }
    else
    {
        this->draw(renderer, _modelViewTransform, flags);
    }

    director->popMatrix(MATRIX_STACK_TYPE::MATRIX_STACK_MODELVIEW);
    
    // FIX ME: Why need to set _orderOfArrival to 0??
    // Please refer to https://github.com/cocos2d/cocos2d-x/pull/6920
    // reset for next frame
    // _orderOfArrival = 0;
}

(1-2-2-1-1-1) 对节点的所有孩子排序,通过调用draw函数,首先绘制ZOrder<0的节点,在绘制自身,最后绘制ZOrder>0的节点. 我们进入draw看看它做些什么. 注意,visit和draw都是虚函数, 以sprite为例,我们进入到/cocos2d-x-3.2/cocos/2d/CCSprite.cpp:

void Sprite::draw(Renderer *renderer, const Mat4 &transform, uint32_t flags)
{
    // Don't do calculate the culling if the transform was not updated
    _insideBounds = (flags & FLAGS_TRANSFORM_DIRTY) ? renderer->checkVisibility(transform, _contentSize) : _insideBounds;

    if(_insideBounds)
    {
        _quadCommand.init(_globalZOrder, _texture->getName(), getGLProgramState(), _blendFunc, &_quad, 1, transform);
        renderer->addCommand(&_quadCommand);
// 物理引擎相关绘制边界
#if CC_SPRITE_DEBUG_DRAW
        _customDebugDrawCommand.init(_globalZOrder);
        _customDebugDrawCommand.func = CC_CALLBACK_0(Sprite::drawDebugData, this);
        renderer->addCommand(&_customDebugDrawCommand);
#endif //CC_SPRITE_DEBUG_DRAW
    }
}

(1-2-2-1-1-1-1) 从代码中可以看出,Sprite的draw函数里面并没有做实际的渲染工作,而是用QuadCommand命令将渲染操作打包,加入到渲染队列里面,在drawscene最后通过调用render()进行统一渲染;我们可以看看_quadCommand.init里面究竟做了什么,找到/cocos2d-x-3.2/cocos/renderer/CCQuadCommand.cpp:

void QuadCommand::init(float globalOrder, GLuint textureID, GLProgramState* glProgramState, BlendFunc blendType, V3F_C4B_T2F_Quad* quad, ssize_t quadCount, const Mat4 &mv)
{
    CCASSERT(glProgramState, "Invalid GLProgramState");
    CCASSERT(glProgramState->getVertexAttribsFlags() == 0, "No custom attributes are supported in QuadCommand");

    _globalOrder = globalOrder;

    _quadsCount = quadCount;
    _quads = quad;
    
    // 设置MV矩阵
    _mv = mv;

    if( _textureID != textureID || _blendType.src != blendType.src || _blendType.dst != blendType.dst || _glProgramState != glProgramState) {
        // 
        _textureID = textureID;
       // _blendType就是我们的BlendFunc混合函数
        _blendType = blendType;
        _glProgramState = glProgramState;
       
        // 生成材质ID
        generateMaterialID();
    }
}

(1-2-2-1-1-1-1-1) 我们在进入到generateMaterialID()函数里面看看:

void QuadCommand::generateMaterialID()
{

    if(_glProgramState->getUniformCount() > 0)
    {
        _materialID = QuadCommand::MATERIAL_ID_DO_NOT_BATCH;
    }
    else
    {
        int glProgram = (int)_glProgramState->getGLProgram()->getProgram();
        int intArray[4] = { glProgram, (int)_textureID, (int)_blendType.src, (int)_blendType.dst};

        _materialID = XXH32((const void*)intArray, sizeof(intArray), 0);
    }
}

从这里我们可以看出, 我们的材质ID（_materialID）最终是要由shader（glProgram）、混合类型（_blendType）、纹理ID（_textureID）组成的, 所以这三样东西如果有谁不一样的话，那就无法生成相同的材质ID，也就无法在同一 个批次里进行渲染了。

(1-2-2-1-2) 现在,我们回到(1-2-2-1-1-1)的draw函数, 通过上面将渲染指令初始化之后,就是将打包好的渲染命令添加到渲染队列里面了.这里只需简单调用renderer->addCommand(&_quadCommand);即可. 这样,(1-2-2-1)处的drawscene函数中,visit通过调用派生类节点添加渲染指令到渲染队列的工作已经完成了.接下来要做的就是做实际的渲染工作了.3.x版本与之前版本不同,是在drawscene最后通过调用render()函数进行统一渲染的,我们进入render()看一下,找到cocos2d-x-3.2/cocos/renderer/CCRenderer.cpp:

void Renderer::render()
{
    //Uncomment this once everything is rendered by new renderer
    //glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);

    //TODO setup camera or MVP
    _isRendering = true;
    
    if (_glViewAssigned)
    {
        // cleanup
        _drawnBatches = _drawnVertices = 0;

        //Process render commands
        //1. Sort render commands based on ID
        for (auto &renderqueue : _renderGroups)
        {
            renderqueue.sort();
        }
        visitRenderQueue(_renderGroups[0]);
        flush();
    }
    clean();
    _isRendering = false;
}

 从代码可以看出,从Cocos2d-x3.0开始，Cocos2d-x引入了新的渲染流程，它不像2.x版本 直接在每一个node中的draw函数中直接调用OpenGL代码进行图形渲染，而是通过各种RenderCommand封装起来，然后添加到一个 CommandQueue队列里面去，而现在draw函数的作用就是在此函数中设置好相对应的RenderCommand参数，然后把此 RenderCommand添加到CommandQueue中。最后在每一帧结束时调用renderer函数进行渲染，在renderer函数中会根据 ID对RenderCommand进行排序，然后才进行渲染。

(1-2-2-1-2-1) 现在我们进入visitRenderQueue函数看看它做了什么动作:

void Renderer::visitRenderQueue(const RenderQueue& queue)
{
    ssize_t size = queue.size();
    
    for (ssize_t index = 0; index < size; ++index)
    {
        auto command = queue[index];
        auto commandType = command->getType();
　　　　if(RenderCommand::Type::QUAD_COMMAND == commandType)
        {
            flush3D();
            auto cmd = static_cast<QuadCommand*>(command);
            //Batch quads
            // 如果Quad数据量超过VBO的大小，那么调用绘制，将缓存的命令全部绘制
            if(_numQuads + cmd->getQuadCount() > VBO_SIZE)
            {
                CCASSERT(cmd->getQuadCount()>= 0 && cmd->getQuadCount() < VBO_SIZE, "VBO is not big enough for quad data, please break the quad data down or use customized render command");
                
                //Draw batched quads if VBO is full
                drawBatchedQuads();
            }
　　　　　　 // 这个处理主要是把命令存入_batchedQuadCommands中，如果如果Quad数据量超过VBO的大小，那么调用绘制，将缓存的命令全部绘制.
　　　　　 　// 如果一直没有超过VBO的大小，drawBatchedQuads绘制函数将在flush被调用时调用.
            // 将命令缓存起来，先不调用绘制
            _batchedQuadCommands.push_back(cmd);
            
            memcpy(_quads + _numQuads, cmd->getQuads(), sizeof(V3F_C4B_T2F_Quad) * cmd->getQuadCount());
            // 通过MV矩阵, 转换成世界坐标
            convertToWorldCoordinates(_quads + _numQuads, cmd->getQuadCount(), cmd->getModelView());
            // 记录下四边形数量
            _numQuads += cmd->getQuadCount();

        }
        else if(RenderCommand::Type::GROUP_COMMAND == commandType)
        {
            flush();
            int renderQueueID = ((GroupCommand*) command)->getRenderQueueID();
            visitRenderQueue(_renderGroups[renderQueueID]);
        }
        else if(RenderCommand::Type::CUSTOM_COMMAND == commandType)
        {
            flush();
            auto cmd = static_cast<CustomCommand*>(command);
            cmd->execute();
        }
        else if(RenderCommand::Type::BATCH_COMMAND == commandType)
        {
            flush();
            auto cmd = static_cast<BatchCommand*>(command);
            cmd->execute();
        }
        else if (RenderCommand::Type::MESH_COMMAND == commandType)
        {
            flush2D();
            auto cmd = static_cast<MeshCommand*>(command);
            if (_lastBatchedMeshCommand == nullptr || _lastBatchedMeshCommand->getMaterialID() != cmd->getMaterialID())
            {
                flush3D();
                cmd->preBatchDraw();
                cmd->batchDraw();
                _lastBatchedMeshCommand = cmd;
            }
            else
            {
                cmd->batchDraw();
            }
        }
        else
        {
            CCLOGERROR("Unknown commands in renderQueue");
        }
    }
}

从代码中，我们看到RenderCommand类型有QUAD_COMMAND,CUSTOM_COMMAND,BATCH_COMMAND,GROUP_COMMAND,MESH_COMMAND五种，OpenGL的API调用是在Renderer::drawBatchedQuads()、BatchCommand::execute()中。通过上面代码的注释,可以看到最常用的QUAD_COMMAND类型的渲染命令的处理过程.

(1-2-2-1-2-1-1) 如果Quad数据量超过VBO的大小(VBO_SIZE = 65536 / 6;), 则会调用drawBatchedQuads进行批量渲染:

void Renderer::drawBatchedQuads()
{
    //TODO we can improve the draw performance by insert material switching command before hand.

    int quadsToDraw = 0;
    int startQuad = 0;

    //Upload buffer to VBO
    if(_numQuads <= 0 || _batchedQuadCommands.empty())
    {
        return;
    }
    // 是否支持VAO
    if (Configuration::getInstance()->supportsShareableVAO())
    {
        //Set VBO data 绑定VBO数据, 激活缓冲区对象
        glBindBuffer(GL_ARRAY_BUFFER, _buffersVBO[0]);

        // option 1: subdata
//        glBufferSubData(GL_ARRAY_BUFFER, sizeof(_quads[0])*start, sizeof(_quads[0]) * n , &_quads[start] );

        // option 2: data
//        glBufferData(GL_ARRAY_BUFFER, sizeof(quads_[0]) * (n-start), &quads_[start], GL_DYNAMIC_DRAW);

        // option 3: orphaning + glMapBuffer
　　　　　// 用数据分配和初始化缓冲区对象
        glBufferData(GL_ARRAY_BUFFER, sizeof(_quads[0]) * (_numQuads), nullptr, GL_DYNAMIC_DRAW);
        // OPENGL 缓冲区对象（buffer object），允许应用程序显式地指定把哪些数据存储在图形服务器或显存中
　　　　 // 返回指向缓冲区的指针, 缓冲一经具体使用之后，只需要改变缓冲区的内容，即在glMapBuffer和glUnmapBuffer之间改变数据即可
        void *buf = glMapBuffer(GL_ARRAY_BUFFER, GL_WRITE_ONLY);
        memcpy(buf, _quads, sizeof(_quads[0])* (_numQuads));
        glUnmapBuffer(GL_ARRAY_BUFFER);
        // 解除绑定
        glBindBuffer(GL_ARRAY_BUFFER, 0);

        //Bind VAO 绑定VAO
        GL::bindVAO(_quadVAO);
    }
    else
    {
#define kQuadSize sizeof(_quads[0].bl)
        glBindBuffer(GL_ARRAY_BUFFER, _buffersVBO[0]);

        glBufferData(GL_ARRAY_BUFFER, sizeof(_quads[0]) * _numQuads , _quads, GL_DYNAMIC_DRAW);

        GL::enableVertexAttribs(GL::VERTEX_ATTRIB_FLAG_POS_COLOR_TEX);

        // vertices
        glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_POSITION, 3, GL_FLOAT, GL_FALSE, kQuadSize, (GLvoid*) offsetof(V3F_C4B_T2F, vertices));

        // colors
        glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_COLOR, 4, GL_UNSIGNED_BYTE, GL_TRUE, kQuadSize, (GLvoid*) offsetof(V3F_C4B_T2F, colors));

        // tex coords
        glVertexAttribPointer(GLProgram::VERTEX_ATTRIB_TEX_COORD, 2, GL_FLOAT, GL_FALSE, kQuadSize, (GLvoid*) offsetof(V3F_C4B_T2F, texCoords));

        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, _buffersVBO[1]);
    }

    //Start drawing verties in batch
    for(const auto& cmd : _batchedQuadCommands)
    {
        auto newMaterialID = cmd->getMaterialID();
        if(_lastMaterialID != newMaterialID || newMaterialID == QuadCommand::MATERIAL_ID_DO_NOT_BATCH)
        {
            //Draw quads
            if(quadsToDraw > 0)
            {
                // 四边形都可以由2个三角形组合而成,指定6个索引点(画出2个GL_TRIANGLES)
                glDrawElements(GL_TRIANGLES, (GLsizei) quadsToDraw*6, GL_UNSIGNED_SHORT, (GLvoid*) (startQuad*6*sizeof(_indices[0])) );
                _drawnBatches++;
                _drawnVertices += quadsToDraw*6;

                startQuad += quadsToDraw;
                quadsToDraw = 0;
            }

            //Use new material
            cmd->useMaterial();
            _lastMaterialID = newMaterialID;
        }

        quadsToDraw += cmd->getQuadCount();
    }

    //Draw any remaining quad
    if(quadsToDraw > 0)
    {
        // 画剩下的四边形
        glDrawElements(GL_TRIANGLES, (GLsizei) quadsToDraw*6, GL_UNSIGNED_SHORT, (GLvoid*) (startQuad*6*sizeof(_indices[0])) );
        _drawnBatches++;
        _drawnVertices += quadsToDraw*6;
    }

    if (Configuration::getInstance()->supportsShareableVAO())
    {
        //Unbind VAO 接除绑定VAO
        GL::bindVAO(0);
    }
    else
    {
        glBindBuffer(GL_ARRAY_BUFFER, 0);
        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
    }

    _batchedQuadCommands.clear();
    _numQuads = 0;
}

附注:5种渲染类型:

1. QUAD_COMMAND：QuadCommand类绘制精灵等。所有绘制图片的命令都会调用到这里，处理这个类型命令的代码就是绘制贴图的openGL代码。
2. CUSTOM_COMMAND：CustomCommand类自定义绘制，自己定义绘制函数，在调用绘制时只需调用已经传进来的回调函数就可以，裁剪节点，绘制图形节点都采用这个绘制，把绘制函数定义在自己的类里。这种类型的绘制命令不会在处理命令的时候调用任何一句openGL代码，而是调用你写好并设置给func的绘制函数。
3. BATCH_COMMAND：BatchCommand类批处理绘制，批处理精灵和粒子,其实它类似于自定义绘制，也不会再render函数中出现任何一句openGL函数。
4. GROUP_COMMAND：GroupCommand类绘制组，一个节点包括两个以上绘制命令的时候，把这个绘制命令存储到另外一个_renderGroups中的元素中，并把这个元素的指针作为一个节点存储到_renderGroups[0]中。

5. MESH_COMMAND :