[webrtc-encoded-transform] SDP mangling: control encoding or control packetization? (#199)

alvestrand has just created a new issue for https://github.com/w3c/webrtc-encoded-transform:

== SDP mangling: control encoding or control packetization? ==
(Unusual usage of bug tracker: asking for input on a feature of an outstanding PR)
While pondering #186 , I wonder if I should redesign.

The purpose of having a packetizer field in the codec description is to control the packetizer/depacketizer once the codec has been negotiated. But putting it in the SDP description kind of mixes the SDP negotiation (outside of transport) with the transport control.

This came to a head when considering how to set the outgoing codec, given that we have a new API for setting it using SetParameters.

New suggestion:
- Drop the packetizer from the negotiation. The two parties have to agree that they are willing to understand what this codec name represent, that's *all* the SDP module should care about.
- Use SetParameters() to get the encoder to encode the type that the transform expects as input. This already exists.
- Add a new call - RtpSender.SetPacketizer(PT, DOMString) - where the MIME type  has a predefined effect, others may be added.
- Similarly, RtpReceiver.SetDepacketizer(PT, DOMString)
- If no packetizer /depacketizer is registered for a PT when a frame/packet is sent for packetization/depacketization, drop it.

It seems cleaner.

@jan-ivar @youennf @orphis @henbos - WDYT?


Please view or discuss this issue at https://github.com/w3c/webrtc-encoded-transform/issues/199 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Thursday, 31 August 2023 13:36:03 UTC