1137472 - VP9 support in signaling and conduit/pipeline code

Status: RESOLVED FIXED

(Core :: WebRTC: Signaling, defect)

Description

[Randell Jesup [:jesup] (needinfo me)]

+++ This bug was initially created as a clone of Bug #1090742 +++

VP9 SDP appears to be the same (or extremely similar to) VP8; note that this likely isn't finalized as the packetization draft for VP9 is far from finished.

This bug covers the non-upstream (webrtc.org) basic portions of vp9 support; it also doesn't cover things about *when* VP9 should be preferred or switched to/from (which may require allowing accepting multiple codecs, which we don't do yet)

Comment 1

[Randell Jesup [:jesup] (needinfo me)]

Attachment: Basic VP9 signaling/pipeline/conduit support

Comment 2

[Randell Jesup [:jesup] (needinfo me)]

Attachment: Basic VP9 signaling/pipeline/conduit support

Now has media.peerconnection.video.vp9_enabled to enable vp9 support

Comment 3

[Byron Campen [:bwc]]

Comment on attachment 8570705 [details] [diff] [review]
Basic VP9 signaling/pipeline/conduit support

Review of attachment 8570705 [details] [diff] [review]:
-----------------------------------------------------------------

Let's add some parse-code testing in media/webrtc/signaling/test/sdp_unittests.cpp for VP9.

I'm also pretty sure we'll want some updates in media/webrtc/signaling/test/jsep_session_unittest.cpp (in ValidateOfferedCodecParams and ValidateAnsweredCodecParams).

Otherwise, just some nits I think.

::: media/webrtc/signaling/src/peerconnection/PeerConnectionImpl.cpp
@@ +914,5 @@
>  
> +  bool vp9Enabled = false;
> +  branch->GetBoolPref("media.peerconnection.video.vp9_enabled",
> +                      &vp9Enabled);
> +  

Some trailing ws here.

::: media/webrtc/signaling/src/sdp/SdpAttribute.h
@@ +1018,1 @@
>    class VP8Parameters : public Parameters

We have a comment to this effect pretty much everywhere this is used, probably would be better to just rename. VP8orVP9Parameters would work for me, I guess VPxParameters might work too.

Comment 4

[Eric Rescorla (:ekr)]

Comment on attachment 8570705 [details] [diff] [review]
Basic VP9 signaling/pipeline/conduit support

Review of attachment 8570705 [details] [diff] [review]:
-----------------------------------------------------------------

Does this interoperate with Chrome's VP9?

Comment 5

[Randell Jesup [:jesup] (needinfo me)]

> Does this interoperate with Chrome's VP9?

Given https://code.google.com/p/webrtc/issues/detail?id=4168 (no VP9 packetization support), interop is rather overrated....  We're using the "Generic" packetization code, and it appears they are also (I fixed a nasty bug in it; I haven't checked if they've fixed the same bug in Canary - I suspect so).

However, tests using talky.io against Canary show working VP9 in both directions, and working regardless of which side offers/accepts.  Note that if Chrome accepts, they accept VP9 and VP8, and so they may try mid-call codec switches (which we'll want to enable depending on load and bitrate too).

Comment 6

[Randell Jesup [:jesup] (needinfo me)]

Attachment: test vp9 sdp in sdp_unittests

Comment 7

[Randell Jesup [:jesup] (needinfo me)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/942cffff9fea

Comment 8

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/942cffff9fea

Comment 9

[Randell Jesup [:jesup] (needinfo me)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/d2f544709115

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/d2f544709115