W3C home > Mailing lists > Public > public-media-capture@w3.org > August 2013

Re: MediaStream Recording : record() method

From: Martin Thomson <martin.thomson@gmail.com>
Date: Wed, 28 Aug 2013 14:15:33 -0700
Message-ID: <CABkgnnXEoQYPU+aD3+OtdtbnV8jav62471sacu6xZstod_Th8g@mail.gmail.com>
To: "Mandyam, Giridhar" <mandyam@quicinc.com>
Cc: Harald Alvestrand <harald@alvestrand.no>, "public-media-capture@w3.org" <public-media-capture@w3.org>
On 28 August 2013 14:01, Mandyam, Giridhar <mandyam@quicinc.com> wrote:
> Thanks for the response, but I am still not understanding the point being made.

The point is that the characteristics of the response to packet loss
are very different.  MediaRecorder isolates the generation of time
slices from the effect of network conditions.  RTCPeerConnection does
not.

The codec choice matters very little in this.  At some point, packet
loss will cause a cap that will manifest as a gap in
RTCPeerConnection.  Likely, TCP transport only increases the amount of
packet loss that is required before the gap appears.

In comparison, MediaRecorder + WebSocket or HTTP upload will probably
not notice the glitch until local buffers blow out.  (If MediaRecorder
is encoding at 4kbps, that's going to be a long time.)  In a sense,
it's a sliding scale, but it's a pretty long way between the two
extremes.
Received on Wednesday, 28 August 2013 21:16:01 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:26:19 UTC