Closing in on WebRTC v1.0

One of the meeting rooms at Microsoft Building 99 in Redmond.

First, thank you Microsoft for hosting the two day W3C WebRTC event. This was the first face to face meeting since the rechartering of the W3C WebRTC working group this summer. The meeting was mostly spent finalising the object model for v1.0 and agreeing on the changes to come. For example, WebRTC v1.0 will have the following objects:

• IceTransport
• DtlsTransport
• SctpTransport
• RtpSender
• RtpReceiver

Broadly, these objects will allow better control of the media pipeline and perhaps avoid munging the Session Description Protocol (SDP). Major discussion highlights were related to:

1) Codec selection: There are two cases of codec selection: (a) before the SDP Offer is sent to the remote party, (b) after multiple codecs are selected. For example, being able to pick H.264 ahead of VP8, when talking to an endpoint or service that supports just one codec. Or being able to pick VP9, if both VP8 and VP9 are successfully negotiated.

2) Quality configuration: being able to set spatial versus temporal bias. For example, screen sharing a slideshare might prefer having higher resolution instead of more frame rate, whereas normal video conferencing (talking heads) might prefer more frame rate instead of higher resolution. This is also a good input for the congestion control algorithm implemented within the browser to decide on what trade-off to make.

3) Connection Setup state: The connection depends on the IceTransport and DtlsTransport, the application needs to introspect both objects to assess the peerconnection’s state. The pc.connectionState, combines the two states and the application needs to query just one object to know if the peerconnection is working or not.

4) IP address privacy: Exposing local addresses which were assumed to be hidden when using a VPN. Chrome and Firefox announced updates, i.e., both browsers have plugins that when installed will filter all but the publicly visible candidates (they use the default ipadress (route) that connected to the service). In FF 43 and Chrome 47, this would be the default behaviour and the plugin will not be useful anymore.

There were many other changes discussed: Audio Levels, unassigned media, replace track, screen sharing, simulcast, etc. The complete slideset and meeting notes are available on the W3C meeting wiki.

If you would like a a personal demo of callstats.io or a free trial coupon email us at support-at-callstats.io.