是否有实现异步游戏引擎循环的最佳实践？

Question

我已经实现了如下游戏引擎循环:

public static Boolean Start ( )
{
    if (hasBoard)
    {
        //  start engine on worker thread
        asyncTask = new AsyncResult ( stopEngine, asyncTask );
        isRunning = ThreadPool.QueueUserWorkItem ( startEngine, asyncTask );

        if (isRunning)
        {
            Console.WriteLine ( "[{0}] Engine started",
                DateTime.Now.ToString ( "hh:mm:ss" ) );
        }
        else
        {
            Console.WriteLine ( "[{0}] Engine failed to start",
                DateTime.Now.ToString ( "hh:mm:ss" ) );
        }
    }

    return isRunning;
}

public static void Stop ( )
{
    Console.WriteLine ( "[{0}] Engine stopping",
        DateTime.Now.ToString ( "hh:mm:ss" ) );

    asyncTask.SetAsCompleted ( null, false );
}

private static void startEngine ( Object task )
{
    while (!( (IAsyncResult)task ).IsCompleted)
    {
        Thread.Sleep ( 10000 );

        Console.WriteLine ( "[{0}] Engine running",
            DateTime.Now.ToString ( "hh:mm:ss" ) );
    }
}

private static void stopEngine ( IAsyncResult iaResult )
{
    //  clean up resources

    Console.WriteLine ( "[{0}] Engine stopped", 
        DateTime.Now.ToString ( "hh:mm:ss" ) );

    isRunning = false;
}

我正在使用AsyncResultJeff Richter在他的文章" 实现CLR异步编程模型"中推荐的类.为了能够从UI停止引擎,我使用的实现与标准异步模式略有不同.这个实现按预期工作,但是当我偏离标准实践时,我回到SO社区以确保我以正确的方式做事.

这个实现有任何问题,任何人都可以看到？

Answer 1

由于这听起来像你可以控制的项目,我建议你抛弃APM并使用Task.NET4中提供的基于模型的模型.这是.NET4而不是APM的推荐方法.该Task班是一部分任务并行库(TPL),但它是伟大的这些基本异步作业为好.

private CancellationTokenSource cts;

public void StartEngine()
{
    if (cts == null)
    {
        cts = new CancellationTokenSource();
        Task.Factory.StartNew(() => GameLoop(cts.Token), cts.Token);
    }
}

private void GameLoop(CancellationToken token)
{
    while (true)
    {
        token.ThrowIfCancellationRequested();
        Thread.Sleep(1000);
        Debug.WriteLine("working...");
    }
}

public void StopEngine()
{
    if (cts != null)
    {
        cts.Cancel();
        cts = null;
    }
}