Bug 1328761 part 2. Update some FetchDriver comments to more clearly reflect reality. r=bkelly

dom/fetch/FetchDriver.cpp

@@ -314,12 +314,22 @@ FetchDriver::HttpFetch()
     nsAutoCString contentType;
     ErrorResult result;
     mRequest->Headers()->GetFirst(NS_LITERAL_CSTRING("content-type"), contentType, result);
-    // This is an error because the Request constructor explicitly extracts and
-    // sets a content-type per spec.
+    // We don't actually expect "result" to have failed here: that only happens
+    // for invalid header names.  But if for some reason it did, just propagate
+    // it out.
     if (result.Failed()) {
       return result.StealNSResult();
     }
 
+    // Now contentType is the header that was set in mRequest->Headers(), or a
+    // void string if no header was set.
+#ifdef DEBUG
+    bool hasContentTypeHeader =
+      mRequest->Headers()->Has(NS_LITERAL_CSTRING("content-type"), result);
+    MOZ_ASSERT(!result.Failed());
+    MOZ_ASSERT_IF(!hasContentTypeHeader, contentType.IsVoid());
+#endif // DEBUG
+
     nsCOMPtr<nsIInputStream> bodyStream;
     mRequest->GetBody(getter_AddRefs(bodyStream));
     if (bodyStream) {