moq-wg / moq-transport

draft-ietf-moq-transport
Other
87 stars 22 forks source link

Header Length for improving parsing #520

Closed suhasHere closed 1 month ago

suhasHere commented 2 months ago

Addresses #474

vasilvv commented 2 months ago

I think this is good for control messages, but I'd mildly prefer to keep the data stream headers without length, since those are sensitive to the framing overhead.

TimEvens commented 2 months ago

I think this is good for control messages, but I'd mildly prefer to keep the data stream headers without length, since those are sensitive to the framing overhead.

Suppose it depends how often streams are changing. Adding object header length does add overhead, but it should be one byte in most uses.

fluffy commented 2 months ago

Victors comment is really making me think about if we need this on objects or not. I'm very torn and think we should talk about how to do while still keeping the framing overhead as low as possible.

afrind commented 2 months ago

Chair Comment:

It seems like there's general support for control messages, but some discussion required for objects. Can we move the object changes to a separate PR so we can land the control changes for the next draft cut?