DownloadFileAsync抛出未处理的异常？

Question

I'm using a WebClient in C# to download a file. I'm using client.DownloadFileAsnyc(). In the past this worked fine, any exceptions would be caught and returned in the completion handler (AsyncCompletedEventArgs.Error)

But now I'm finding that if I run out of space in the destination location during the download, the IOExcption is getting thrown and causing the app to crash. Does anyone know why this wouldn't be caught and returned in the completion handler?
Note: I've also tried putting the DownloadFileAsync line in a try catch. Still no go

Here's the code:

_client = new WebClient();
_client.DownloadProgressChanged += ProgressChanged;
_client.DownloadFileCompleted += DownloadComplete;
_client.DownloadFileAsync(new Uri(url), Destination);

private void DownloadComplete(object sender, AsyncCompletedEventArgs args)
{
}

This is being compiled under .NET 3.5.

Answer 1

The MSDN docs for System.ComponentModel indicate that your DownloadFileCompleted handler should indeed receive the exception, but apparently that isn't happening here. You might try hooking some of the other *Completed events on WebClient to see if it is getting sent there.

In any case, you aren't catching the exception because it isn't happening on the thread that executes the try / catch block.

When you use the async api functions (which generally means any function name that ends with "Async" or "Begin"), the actual operation happens in the threadpool, not on the thread where you launched the operation. Surrounding the operation in a try/catch will not catch failures on the background threads.

To properly catch all exceptions in your application, you can install a global exception handler that will be called whenever an exception is thrown that isn't caught somewhere else in your program.

One easy way around this behavior would be to use the synchronous client.DownloadFile() function, and then call that function from a background thread so you don't block the main thread of your program. Here is a quick-and-dirty example that demonstrates it:

// create a thread function to download a file synchronously
function DoDownload(object state){
    List<String> st = (List<String>)(state);
    String uri = st[0];
    String fname = st[1];
    try {
        client.DownloadFile(uri, fname);
    } catch {
        // you'll catch the exception here because the 
        // try/catch is on the same thread that is doing the downloading
    }
}

//  somewhere else, when you want to actually start the download:
String uri = "http://example.com/the_file.txt";
string theFileName = "destination.txt";
List<String> st = new List<String>{ theUri, theFileName };
ThreadPool.QueueUserWorkItem(DoDownload, st);

Please note, this example is a bit of an abuse of the system threadpool, especially if you are using it for large files where the download will take more than a second or so. If you are downloading larger files, or are downloading many files simultaneously, you definitely should not do it this way.