Docs 0.11.0

#pragma once


#include "ofMain.h"
#include <atomic>

/// This is a simple example of a ThreadedObject created by extending ofThread.
/// It contains data (count) that will be accessed from within and outside the
/// thread and demonstrates several of the data protection mechanisms (aka
/// mutexes).
class ThreadedObject: public ofThread
{
public:
    /// On destruction wait for the thread to finish
    /// so we don't destroy the pixels while they are
    /// being used. Otherwise we would crash
    ~ThreadedObject(){
        stop();
        waitForThread(false);
    }

    void setup(){
        pixels.allocate(640,480,OF_PIXELS_GRAY);
        tex.allocate(pixels);
        start();
    }

    /// Start the thread.
    void start(){
        startThread();
    }

    /// Signal the thread to stop.  After calling this method,
    /// isThreadRunning() will return false and the while loop will stop
    /// next time it has the chance to.
    /// In order for the thread to actually go out of the while loop
    /// we need to notify the condition, otherwise the thread will
    /// sleep there forever.
    /// We also lock the mutex so the notify_all call only happens
    /// once the thread is waiting. We lock the mutex during the
    /// whole while loop but when we call condition.wait, that
    /// unlocks the mutex which ensures that we'll only call
    /// stop and notify here once the condition is waiting
    void stop(){
        std::unique_lock<std::mutex> lck(mutex);
        stopThread();
        condition.notify_all();
    }

    /// Everything in this function will happen in a different
    /// thread which leaves the main thread completelty free for
    /// other tasks.
    void threadedFunction(){
        while(isThreadRunning()){
			// since we are only writting to the frame number from one thread
			// and there's no calculations that depend on it we can just write to
			// it without locking
			threadFrameNum++;

			// Lock the mutex until the end of the block, until the closing }
			// in which this variable is contained or we unlock it explicitly
            std::unique_lock<std::mutex> lock(mutex);

            // The mutex is now locked so we can modify
            // the shared memory without problem
            auto t = ofGetElapsedTimef();
            for(auto line: pixels.getLines()){
                auto x = 0;
                for(auto pixel: line.getPixels()){
                    auto ux = x/float(pixels.getWidth());
                    auto uy = line.getLineNum()/float(pixels.getHeight());
                    pixel[0] = ofNoise(ux, uy, t);
                    x++;
                }
            }

            // Now we wait for the main thread to finish
            // with this frame until we can generate a new one
            // This sleeps the thread until the condition is signaled
            // and unlocks the mutex so the main thread can lock it
            condition.wait(lock);
        }
    }

    void update(){
        // if we didn't lock here we would see
        // tearing as the thread would be updating
        // the pixels while we upload them to the texture
		std::unique_lock<std::mutex> lock(mutex);
        tex.loadData(pixels);
        condition.notify_all();
    }

	void updateNoLock(){
		// we don't lock here so we will see
		// tearing as the thread will update
		// the pixels while we upload them to the texture
		tex.loadData(pixels);
		condition.notify_all();
	}

    /// This drawing function cannot be called from the thread itself because
    /// it includes OpenGL calls
    void draw(){
        tex.draw(0,0);
    }

	int getThreadFrameNum(){
		return threadFrameNum;
	}

protected:
    // pixels represents shared data that we aim to always access from both the
    // main thread AND this threaded object and at least from one of them for
    // writing. Therefore, we need to protect it with the mutex.
    // Otherwise it wouldn't make sense to lock.
    ofFloatPixels pixels;

    ofTexture tex;
    std::condition_variable condition;
	int threadFrameNum = 0;
};