WebRtcVoiceEngine: Use AudioDecoderFactory to generate recv codecs.

webrtc/modules/audio_coding/codecs/audio_format.h

Index: webrtc/modules/audio_coding/codecs/audio_format.h
diff --git a/webrtc/modules/audio_coding/codecs/audio_format.h b/webrtc/modules/audio_coding/codecs/audio_format.h
index 61c0dd9f6fa127c32c3fd075c71c10b993419ff9..09075e49acba9b378af2aa47bc993cd9fc9181a2 100644
--- a/webrtc/modules/audio_coding/codecs/audio_format.h
+++ b/webrtc/modules/audio_coding/codecs/audio_format.h
@@ -37,6 +37,7 @@ struct SdpAudioFormat {
 
   SdpAudioFormat& operator=(const SdpAudioFormat&);
   SdpAudioFormat& operator=(SdpAudioFormat&&);
+  bool operator<(const SdpAudioFormat& b) const;

[ossu, 2016/06/16 16:21:24]

So we can put them in a map/set.

[ivoc, 2016/07/06 15:24:51]

Do they need to be sorted? If not, how about unordered_map/unordered_set
instead? 

[ossu, 2016/07/06 16:23:25]

For unordered_map, we'd need to define a hasher instead, I believe. Having
operator< here allows us to put it in sets or maps, which is generally a nice
thing to be able to do, and generally what's done. Only a handful of places in
our code use unordered_map.

[ivoc, 2016/07/08 12:50:32]

Oh right, I didn't realize that, I guess I learned something new :-).

 
   std::string name;
   int clockrate_hz;