Re: [webrtc-pc] Clarify the definition of "playout" for `RTCRtpContributingSource`. (#2172)

How about we do this:

1. I'll first create a PR to make it so that each `RTCRtpContributingSource` SHOULD be recorded when a frame from a RTP packet is delivered to `MediaStreamTrack`. `timestamp` would then be defined in terms of the instant of that delivery. Firefox's current implementation doesn't do the SHOULD, but its `timestamp` is still spec-compliant under the "may be implemented in any manner so long as the end result is equivalent" rule.

2. I'll afterwards create a second issue+PR to add a separate `playoutTimestamp`. We could give it more vague requirements:

> The estimated timestamp of when the frame, indicated by `timestamp`, was (or will be) physically played out. Its reference clock is the same as the reference clock for `timestamp`. The `playoutTimestamp` SHOULD include delays from sound card buffers and rendering, if that information is available. It MAY point to an instant in the future. This member MAY be absent if the user agent doesn't have a better estimate than the value stored in `timestamp`.

GitHub Notification of comment by chxg
Please view or discuss this issue at using your GitHub account

Received on Friday, 26 April 2019 10:23:44 UTC