Why does Resharper give me this Constraint Violation, and how can I appease it?

Question

I get several "Possible 'null' assignment to entity marked with 'NotNull' attribute" fingerwags from R# at code such as:

if ((webResponse.StatusCode == HttpStatusCode.OK) && (webResponse.ContentLength > 0))
{
    var reader = new StreamReader(webResponse.GetResponseStream()); // <-- R# hates this line of code

How could the StreamReader be null in this case? If it can, how can I defensively program around that possibility?

UPDATE

Okay, If I change this code:

String strResult;
WebRequest objRequest = WebRequest.Create(strURL);
WebResponse objResponse = objRequest.GetResponse();
using (var sr = new StreamReader(objResponse.GetResponseStream()))
{
    strResult = sr.ReadToEnd();
    sr.Close();
}
return strResult;

...to this:

String strResult = string.Empty;
WebRequest objRequest = WebRequest.Create(strURL);
WebResponse objResponse = objRequest.GetResponse();
using (var sr = new StreamReader(objResponse.GetResponseStream()))
{
    if (sr != null)
    {
        strResult = sr.ReadToEnd();
        sr.Close();
    }
}
return strResult;

...I still get the same fingerwag on the same ("using") line.

And if I change it to this:

String strResult = string.Empty;
WebRequest objRequest = WebRequest.Create(strURL);
WebResponse objResponse = objRequest.GetResponse();
if (objResponse != null)
{
    using (var sr = new StreamReader(objResponse.GetResponseStream()))
    {
        strResult = sr.ReadToEnd();
        sr.Close();
    }
}
return strResult;

...the "Possible 'null' assignment to entity marked with 'NotNull' attribute" still refers to that same line. So how can I nuke R#'s fingerwag? ISTM that the "using" would wrap up the necessary checking for null...

UPDATE 2

Resharper complains on this line:

var reader = new StreamReader(webResponse.GetResponseStream());

..."Possible 'null' assignment to entity marked with 'NotNull' attribute"

So I changed it to check both webResponse and reader:

var webResponse = (HttpWebResponse)webRequest.GetResponse();
if ((webResponse != null) && (webResponse.StatusCode == HttpStatusCode.OK) && (webResponse.ContentLength > 0))
{
    var reader = new StreamReader(webResponse.GetResponseStream());
    if (reader != null)
    {

...but then I get, "Expression is always true" for both of those "!=" tests; so it seems like it's saying, "Watch out! webResponse could be null!" and/or, "Watch out! reader could be null!" but then saying, "That's wasted code - they will never be null."

Answer 1

if ((webResponse.StatusCode == HttpStatusCode.OK) && (webResponse.ContentLength > 0))
{
    var responsestream = webResponse.GetResponseStream();
    if(responsestream != null) {
        var reader = new StreamReader(responsestream );
    }
}

It's not the StreamReader that's null, it's the ReponseStream. Simply add a check to make sure the returned ResponseStream is not null if you want the warning gone.

Answer 2

It's not the streamreader that might be null, it's the result of webResponse.GetResponseStream()