- From: Varun Singh <vsingh.ietf@gmail.com>
- Date: Fri, 25 Oct 2013 11:27:16 +0300
- To: Harald Alvestrand <harald@alvestrand.no>
- Cc: "public-webrtc@w3.org" <public-webrtc@w3.org>, "Rachel Huang(yihong)" <rachel.huang@huawei.com>, Ron Even <ron.even.tlv@gmail.com>, "Romascanu, Dan (Dan)" <dromasca@avaya.com>
Hi Harald, In http://www.w3.org/2011/04/webrtc/wiki/Stats, I notice that RTCInboundRTPStreamStats does not have fractional loss, was it removed and put some place else? Another clarification: is the RTCInboundRTPStreamStats calculated every time the stats API is called or does it return value from the last RTCP RR? Lastly, we would like to get comments on additional statistics for RTCInboundRTPStreamStats [1] We make a recommendation for 7 additional metrics that can be used to measure goodput, quality of service. Additionally, if anyone wants to review the whole set of metrics available in RTCP XRs, see [2]. Thanks very much, feedback is appreciated. Varun [1] http://tools.ietf.org/html/draft-singh-xrblock-webrtc-additional-stats-01 [2] http://tools.ietf.org/html/draft-huang-xrblock-rtcweb-rtcp-xr-metrics-02 On Fri, Oct 25, 2013 at 10:01 AM, Harald Alvestrand <harald@alvestrand.no> wrote: > I've (finally) made the correction to DOMHighResTimeStamp. > > I also removed firs (because I'm unsure if that's what we want to count) and > volume (because I don't think it's a well defined metric), and moved > priority from candidatepair to candidate, per discussion. > > > On 10/24/2013 07:12 PM, Jan-Ivar Bruaroey wrote: > > Great thanks! - long long for bytes makes sense. I can still get it in for > FF27 (Monday). > > A couple of nits: >> DOMHiResTimeStamp timestamp; > > I think you mean DOMHighResTimeStamp right? > >> unsigned long firs; > > I'm not sure what this is. > > .: Jan-Ivar :. > > On 10/24/13 2:21 AM, Harald Alvestrand wrote: > > I've updated http://www.w3.org/2011/04/webrtc/wiki/Stats with most of the > suggestions, and a few other fiddles. It should also be easy (easier?) now > to generate WebIDL from the source; just take the content of all the <pre> > sections. > > I also made all the byte counters "long long"; we've found that 32-bit byte > counters overflow rapidly. > > Thanks very much for the feedback! > > > -- http://www.netlab.tkk.fi/~varun/
Received on Friday, 25 October 2013 08:40:18 UTC