RTCRtpScriptTransform frame vs packet

[k-wasniowski]

Current proposal of w3c suggests that RTCRtpScriptTransform should be able to handle SFrameTransform as a part of streams readable/writeable as :

onrtctransform = transformer => {
  const decrypter = new SFrameDecryterStream({mode: `packet`);
  transformer.readable
    .pipeThrough(new TransformStream(myMetadataTransform))
    .pipeThrough(decrypter)
    .pipeTo(writable);
};

In theory it seems that this usecase seems valid, however it leads to combination of frame/packet transformers in a single stream. Which causes blink level integration to be hard to achieve.

[k-wasniowski]

Created a related issue here :
w3c/webrtc-encoded-transform#290