Now run EndToEndTest with the WebRTC-NewVideoJitterBuffer experiment.

webrtc/modules/video_coding/rtp_frame_reference_finder.cc

Index: webrtc/modules/video_coding/rtp_frame_reference_finder.cc
diff --git a/webrtc/modules/video_coding/rtp_frame_reference_finder.cc b/webrtc/modules/video_coding/rtp_frame_reference_finder.cc
index 5a9420a84776e168834ae2f69f8a2ec5e4d7c188..5f5c44cf32a1c3faa01e824fe847ad05d53da9e5 100644
--- a/webrtc/modules/video_coding/rtp_frame_reference_finder.cc
+++ b/webrtc/modules/video_coding/rtp_frame_reference_finder.cc
@@ -43,7 +43,6 @@ void RtpFrameReferenceFinder::ManageFrame(
     case kVideoCodecFlexfec:
     case kVideoCodecULPFEC:
     case kVideoCodecRED:
-    case kVideoCodecUnknown:
       RTC_NOTREACHED();
       break;
     case kVideoCodecVP8:
@@ -52,6 +51,7 @@ void RtpFrameReferenceFinder::ManageFrame(
     case kVideoCodecVP9:
       ManageFrameVp9(std::move(frame));
       break;
+    case kVideoCodecUnknown:

[sprang_webrtc, 2016/11/21 16:49:14]

Isn't unknown separate from generic? Can we handle it?

[philipel, 2016/11/22 14:56:23]

I treat it as generic since that is what is used with the current jitter buffer
in the end to end tests. Added a todo to take a look at this.

     case kVideoCodecH264:
     case kVideoCodecI420:
     case kVideoCodecGeneric: