Re: [Wish] Implementation report for draft-ietf-wish-whip-00

[Lorenzo Miniero <[e-mail]>]

On Fri, 10 Sep 2021 11:01:44 +0200
Juliusz Chroboczek <[e-mail]> wrote:

> > Sergio pointed this out too in a private conversation, as the
> > format of the candidates to send in a PATCH was one of the
> > assumptions I made in my implementation. I chose to serialize them
> > to a JSON object exactly because of what you said (sdpMLineIndex),
> > but if bundling is indeed mandated, then it's unneeded and the
> > "raw" candidate is fine. I'll fix this in both server and client
> > tomorrow.  
> 
> Even though we don't need the sdpMLineIndex (due to bundling), we do
> need the ufrag, so simply sending the candidate line won't do.
> 
> Not sure if we need the ice-passwd.  I hope we don't, since it's not
> part of the RTCIceCandidate, so it would need parsing of the SDP on
> the client side.
> 
> -- Juliusz
> 

Yes, I checked RFC 8840 (my fault for missing it initially) and
ice-ugrag, ice-pwd and the "fake" m-line are all required. I'm updating
the code of both client and server right now to adhere to the spec.

Lorenzo

-- 
I'm getting older but, unlike whisky, I'm not getting any better
https://twitter.com/elminiero