Make cricket::VideoCapturer implement VideoSourceInterface

talk/media/base/capturerenderadapter.h

 /*
  * libjingle
  * Copyright 2012 Google Inc.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
  *
  *  1. Redistributions of source code must retain the above copyright notice,
  *     this list of conditions and the following disclaimer.
  *  2. Redistributions in binary form must reproduce the above copyright notice,
@@ skipping 20 matching lines @@
 // CaptureRenderAdapter is Thread-unsafe. This means that none of its APIs may
 // be called concurrently.
 
 #ifndef TALK_MEDIA_BASE_CAPTURERENDERADAPTER_H_
 #define TALK_MEDIA_BASE_CAPTURERENDERADAPTER_H_
 
 #include <vector>
 
 #include "talk/media/base/videocapturer.h"
 #include "webrtc/base/criticalsection.h"
-#include "webrtc/base/sigslot.h"
 #include "webrtc/media/base/videosinkinterface.h"
 
 namespace cricket {
 
 class VideoCapturer;
 class VideoProcessor;
 
-class CaptureRenderAdapter : public sigslot::has_slots<> {
+class CaptureRenderAdapter : public rtc::VideoSinkInterface<VideoFrame> {

[nisse-webrtc, 2016/02/03 09:16:34]

After these changes, it looks like this class is 

1. Functionally equivalent to a VideoTrack without the muting logic. I.e, a
plain mux.

2. Redundant, any code calling AddSink on this object could just as well call
AddSink directly on the VideoCapturer.

Correct?

[pthatcher1, 2016/02/03 15:38:35]

I'd like it if we can just remove this class altogether.   It's only used in one
place:  VideoCapturerState, which only uses it for 2 things:

1.  To help implement CaptureManager::AddVideoSink and
CaptureManager::RemoveVideoSink
2.  To store a video capturer to be used in
CaptureManager::UnregisterVideoCapturer.

For #1, it could probably use our VideoMux (whatever we end up calling it)
instead of CaptureRendererAdapter.  Our we could go one step further: 
CaptureManager::AddVideoSink and CaptureManager::RemoveVideoSink are only used
to implement VideoSource::AddSink and VideoSource::RemoveSink, so maybe we could
just skip using the CaptureManager at all for this.

For #2, the VideoCapturerState could just contain the VideoCapturer directly
instead of having the CaptureRendererAdapter do it.

[perkj_webrtc, 2016/02/08 14:32:00]

I think this can be removed later yes. But since this is currently used I don't
want to grow this cl further. 

  public:
   static CaptureRenderAdapter* Create(VideoCapturer* video_capturer);
   ~CaptureRenderAdapter();
 
   void AddSink(rtc::VideoSinkInterface<VideoFrame>* sink);
   void RemoveSink(rtc::VideoSinkInterface<VideoFrame>* sink);
 
   VideoCapturer* video_capturer() { return video_capturer_; }
+
  private:
-
   explicit CaptureRenderAdapter(VideoCapturer* video_capturer);
   void Init();
 
   // Callback for frames received from the capturer.
-  void OnVideoFrame(VideoCapturer* capturer, const VideoFrame* video_frame);
+  void OnFrame(const VideoFrame& frame) override;
 
   // Just pointers since ownership is not handed over to this class.
   std::vector<rtc::VideoSinkInterface<VideoFrame>*> sinks_;
   VideoCapturer* video_capturer_;
   // Critical section synchronizing the capture thread.
   rtc::CriticalSection capture_crit_;
 };
 
 }  // namespace cricket
 
 #endif  // TALK_MEDIA_BASE_CAPTURERENDERADAPTER_H_