Re: [AVTCORE] Adopt draft-petithuguenin-avtcore-rfc5764-mux-fixes as WG draft?

[Simon Perreault <[e-mail]>]

Le 2015-03-11 21:46, Mo Zanaty (mzanaty) a écrit :
> Do we really need 1280 STUN methods when only 13 are currently defined?
> Do we really need 44 DTLS content types when only 5 are currently defined?
> Do we really need 16384 TURN channels per client?

Yes that's a discussion we need to have.

One thing to keep in mind is that the protocols individually define
ranges that are much larger than what was in RFC 5764, and they do
overlap in some cases. So we end up with codepoints that may be used on
a multiplexed connection and others that may not. IIRC a few
non-multiplexable codepoints have already been assigned. I wouldn't want
to have to define multiplexable and non-multiplexable versions of the
same codepoint because we decided to restrict the range of multiplexable
codepoints.

Summary:
- Ok in principle to restrict the multiplexable codepoint ranges.
- Need to carefully review already-allocated codepoints and ensure we're
not shooting ourselves in the collective foot.

Another thing is that we're currently doing STUN-bis and TURN-bis in
TRAM, so it is a great opportunity to fix this for STUN/TURN at the
source. That is, we could explicitly define multiplexable and
non-multiplexable ranges and sync with this draft.

Simon