Question

I am looking for advice as to how to handle any exceptions thrown in the following code example:

private string SendRequest() {         HttpWebRequest request = (HttpWebRequest)WebRequest.Create(myURL);         // Code initialising HttpWebRequest         HttpWebResponse response = (HttpWebResponse)request.GetResponse();         Stream rcvdStream = response.GetResponseStream();         StreamReader readStream = new StreamReader(rcvdStream, Encoding.UTF8);         string responseString = readStream.ReadToEnd();         response.Close();         readStream.Close();         return responseString; } 

My main concern is to ensure the StreamReader and HttpRequest object are closed whent the method ends. Should I:

1. Wrap the lot up in a try/catch/finally logging any exceptions in the catch block and closing the stream in the finally block?
2. Use a using statement on the HttpWebRequest object instantiation and a nested using statement when creating the StreamReader?
3. Not worry about it and assume GC will clear everything up when the objects go out of scope as the method is exited?

EDIT: Further investigation has revealed that option 2 can be done without nesting the using statements:

    using (HttpWebResponse response = (HttpWebResponse)request.GetResponse())     using (Stream rcvdStream = response.GetResponseStream())     {         StreamReader readStream = new StreamReader(rcvdStream, Encoding.UTF8);         payResponse = readStream.ReadToEnd();     } 

This produces the following IL code which demonstrates that it is effectively creating a nested try/finally block:

  IL_00b0:  callvirt   instance class [System]System.Net.WebResponse [System]System.Net.WebRequest::GetResponse()   IL_00b5:  castclass  [System]System.Net.HttpWebResponse   IL_00ba:  stloc.s    response   .try   {     IL_00bc:  ldloc.s    response     IL_00be:  callvirt   instance class [mscorlib]System.IO.Stream [System]System.Net.WebResponse::GetResponseStream()     IL_00c3:  stloc.s    rcvdStream     .try     {       IL_00c5:  nop       IL_00c6:  ldloc.s    rcvdStream       IL_00c8:  call       class [mscorlib]System.Text.Encoding [mscorlib]System.Text.Encoding::get_UTF8()       IL_00cd:  newobj     instance void [mscorlib]System.IO.StreamReader::.ctor(class [mscorlib]System.IO.Stream,                                                                                  class [mscorlib]System.Text.Encoding)       IL_00d2:  stloc.s    readStream       IL_00d4:  ldloc.s    readStream       IL_00d6:  callvirt   instance string [mscorlib]System.IO.TextReader::ReadToEnd()       IL_00db:  stloc.3       IL_00dc:  nop       IL_00dd:  leave.s    IL_00f3     }  // end .try     finally     {       IL_00df:  ldloc.s    rcvdStream       IL_00e1:  ldnull       IL_00e2:  ceq       IL_00e4:  stloc.s    CS$4$0001       IL_00e6:  ldloc.s    CS$4$0001       IL_00e8:  brtrue.s   IL_00f2       IL_00ea:  ldloc.s    rcvdStream       IL_00ec:  callvirt   instance void [mscorlib]System.IDisposable::Dispose()       IL_00f1:  nop       IL_00f2:  endfinally     }  // end handler     IL_00f3:  nop     IL_00f4:  leave.s    IL_010a   }  // end .try   finally   {     IL_00f6:  ldloc.s    response     IL_00f8:  ldnull     IL_00f9:  ceq     IL_00fb:  stloc.s    CS$4$0001     IL_00fd:  ldloc.s    CS$4$0001     IL_00ff:  brtrue.s   IL_0109     IL_0101:  ldloc.s    response     IL_0103:  callvirt   instance void [mscorlib]System.IDisposable::Dispose()     IL_0108:  nop     IL_0109:  endfinally   }  // end handler 

Answer 1

Options one and two are your best bets. Option 3 IS NOT a good idea.

I am personally quite fond of the Using statement route. However, the try/catch/finally route is essentially the same, and provides you your needed logging mechanism.