Stop using some scoped_ptr features that unique_ptr doesn't have

webrtc/base/asyncudpsocket.cc

Index: webrtc/base/asyncudpsocket.cc
diff --git a/webrtc/base/asyncudpsocket.cc b/webrtc/base/asyncudpsocket.cc
index 51a8fa0af0a2a0cbd4126cba9ffa5647b1c32d15..03f3ab970d942d290ab0845acf6d140e0e0e14cf 100644
--- a/webrtc/base/asyncudpsocket.cc
+++ b/webrtc/base/asyncudpsocket.cc
@@ -37,7 +37,7 @@ AsyncUDPSocket* AsyncUDPSocket::Create(SocketFactory* factory,
 
 AsyncUDPSocket::AsyncUDPSocket(AsyncSocket* socket)
     : socket_(socket) {
-  ASSERT(socket_);
+  ASSERT(!!socket_);

[tommi, 2016/03/15 09:22:09]

The double not is not always not so readable, except that here it is not.
Compare to nullptr instead or even just check socket? (no underscore) 
Alternatively fix ctor to use unique_ptr.

Actually, the pointer is dereferenced below, always, so this isn't necessary. 

[kwiberg-webrtc, 2016/03/15 09:59:41]

OK, will remove. (!!x is the standard C idiom for converting x to a bool---it's
only confusing the first time you see it. I guess in C++ it's more customary to
use bool(x) or static_cast<bool>(x)...)

[tommi (sloooow) - chröme, 2016/03/15 16:49:00]

This isn't the first time I'm seeing it :)

[kwiberg-webrtc, 2016/03/15 18:53:37]

And yet you acted confused...

[tommi, 2016/03/17 12:23:26]

I guess I was rhetorically confused :)

   size_ = BUF_SIZE;
   buf_ = new char[size_];