- From: Philipp Hancke via GitHub <sysbot+gh@w3.org>
- Date: Mon, 19 Dec 2016 21:05:23 +0000
- To: public-webrtc-logs@w3.org
I typed some notes earlier while on the plane: what is the idea behind the selector argument to getStats()? The optional selector argument to getStats() is a MediaStreamTrack. It might imply that the caller is only interested in information about that track. For audiotracks, the caller might look for information about audio levels at a very high rate. Calling getStats and then iterating the resulting map to find the report with the track id might be considered too heavy in such cases. Also it would signal the intent "I only care about this track" to the browser which would allow it to gather/update less information. So lets assume we had an implementation of this selector that would only return the matching RTCMediaStreamTrackStats. What would the ssrcIds be set to? (assuming #85 is fixed) Would they point to an RTCRTPStreamStats object? If so, I assume that this object has a transportId which points to an RTCTransportStats. Which points to a selected candidate pair. In the end, the call with the selector argument will have to return the complete set of statistics. So what is the point? Is this simply an artifact of a previous statistics model that was not updated? I can see this being marginally useful if the callback was something like this: ``` pc.getStats(track) .then((reportForTrack, snowball) => { // look up *Id in snowball }) ``` which would save the initial iteration to find the right starting point. -- GitHub Notification of comment by fippo Please view or discuss this issue at https://github.com/w3c/webrtc-stats/issues/116#issuecomment-268078140 using your GitHub account
Received on Monday, 19 December 2016 21:05:30 UTC