Question

对于GZipStream使用以下构造，当GZipStream是CopyToAsync的目标时，它似乎永远不会调用我的自定义流的* Async方法。

using (var fs = new System.IO.FileStream(@"C:\BTR\Source\Assemblies\BTR.Rbl.Evolution.Documents.dll", 
        System.IO.FileMode.Open, System.IO.FileAccess.Read, System.IO.FileShare.None, 8192, true))
{
    using (var ss = new GZipStream(new MyCustomStream(), CompressionMode.Compress))
    {
        await fs.CopyToAsync(ss);
    }
}

似乎只调用BeginWrite/EndWrite机制。有没有办法从GZipStream派生来调用WriteAsync，以便我的自定义流不必同时实现WriteAsync方法和BeginWrite/EndWrite方法？

您可以找到此here

的工作样本

更新：调用初始Write()方法时调用

SampleStream.Write(buffer, offset, count)
System.IO.Compression.DeflateStream.DoMaintenance(array, offset, count)
System.IO.Compression.DeflateStream.InternalWrite(array, offset, count, isAsync)
System.Runtime.Remoting.Messaging.StackBuilderSink.AsyncProcessMessage(msg, replySink)
System.Runtime.Remoting.Proxies.AgileAsyncWorkerItem.ThreadPoolCallBack(o)
System.Threading.QueueUserWorkItemCallback.WaitCallback_Context(state)
System.Threading.ExecutionContext.RunInternal(executionContext, callback, state, preserveSyncCtx)
System.Threading.ExecutionContext.Run(executionContext, callback, state, preserveSyncCtx)
System.Threading.QueueUserWorkItemCallback.System.Threading.IThreadPoolWorkItem.ExecuteWorkItem()
System.Threading._ThreadPoolWaitCallback.PerformWaitCallback()
(Unmanaged code)

Answer 1

自定义流实施BeginWrite / EndWrite（以及BeginRead / EndRead）更为正确。如果您使用我的AsyncEx.Tasks library：

，这并不难

public override IAsyncResult BeginWrite(byte[] buffer, int offset, int count,
    AsyncCallback callback, object state)
{
  var task = WriteAsync(buffer, offset, count);
  return ApmAsyncFactory.ToBegin(task, callback, state);
}

public override void EndWrite(IAsyncResult asyncResult)
{
  ApmAsyncFactory.ToEnd(asyncResult);
}

（ApmAsyncFactory已添加到1.2.0-alpha-01的{{1}}版本中。

当CopyToAsync的目标时，GZipStream不调用底层流*异步方法

1 个答案: