问题
I am working on solution which uses asynchronous memory streaming and I am thinking about right approach for implementing such. Which one is more convenient? The first, simple:
//First approach: linear async
private async static Task WriteToStreamFirstVariant()
{
MemoryStream memoryStream = new MemoryStream();
byte[] data = new byte[256];
try
{
await memoryStream.WriteAsync(data, 0, data.Length);
}
catch(Exception exception)
{
//Handling exception
}
finally
{
memoryStream.Dispose();
}
}
Or the second with nested tasks and closures?
//Second approach: nested tasks async
private async static Task WriteToStreamSecondVariant()
{
await Task.Run(async () =>
{
byte[] data = new byte[256];
using (MemoryStream memoryStream = new MemoryStream())
{
await memoryStream.WriteAsync(data, 0, data.Length)
.ContinueWith((Task writingTask) =>
{
//Handling exceptions
AggregateException exceptions = writingTask.Exception;
});
}
});
}
回答1:
The fact is that MemoryStream
's Read/WriteAsync
methods don't actually provide any kind of true asynchronous implementation. All they do is perform the operation synchronously and return an already completed Task
. Therefore there is no benefit to calling the async methods when you know it's a MemoryStream
. In fact, it's just completely unnecessary overhead.
Now, forgetting that for a second just to answer your question on style, the first approach is better one because you don't allocate/schedule a new Task
unnecessarily (e.g. with Task::Run
), but I don't know why you wouldn't just use a using()
statement in that approach. So here's the cleanest/simplest IMHO:
private async static Task WriteToStreamFirstVariantSimplified()
{
using(MemoryStream memoryStream = new MemoryStream())
{
byte[] data = new byte[256];
try
{
await memoryStream.WriteAsync(data, 0, data.Length);
}
catch(Exception exception)
{
//Handling exception
}
}
}
来源:https://stackoverflow.com/questions/20805396/asynchronous-memory-streaming-approach-which-of-the-following