[Standards] [Fwd: [MMUSIC] ICE tcp diffs]
stpeter at stpeter.im
Wed Nov 28 19:06:12 UTC 2007
This makes me think that it would be best for XEP-0176 to specify only
the Jingle ice-udp transport, and to move the Jingle ice-tcp transport
definition to a separate XEP.
-------- Original Message --------
Date: Wed, 28 Nov 2007 13:07:20 -0500
From: Jonathan Rosenberg <jdrosen at cisco.com>
To: IETF MMUSIC WG <mmusic at ietf.org>
Subject: [MMUSIC] ICE tcp diffs
As folks may have noticed, and updated ICE-tcp is now available:
the changes from -04 are:
* fixed bugs in pseudocode
SO_REUSEPORT -> SO_REUSEADDR
for loop needs to go from 0 to MAX, not 1 to MAX
* clarified that secure media is always DTLS-SRTP, even over a TCP
connection. Forbid RTP over TLS.
* added checks on sender side to make sure application data doesn't
match a stun packet, and if it does, suggests to use a different
number of bytes to make the data not look like a stun message based
on the length checks
* previously, if a clients tcp connection to a turn server failed
during the middle of a session, the client would reconnect and
perform a 'move'. Moves have been removed from turn so now the draft
says to reconnect and restart ICE for the stream
I did not add ice-lite to ICE-tcp; I started on it to see how it would
look, but ran out of time before finishing. That is the main issue I
want to discuss in Vancouver.
The other thing to note is that ICE-tcp depends on TURN-tcp, which is a
new draft that has been split off from the main TURN spec, and will move
at a slower pace (and only once the core UDP TURN stuff is done).
Jonathan D. Rosenberg, Ph.D. 499 Thornall St.
Cisco Fellow Edison, NJ 08837
Cisco, Voice Technology Group
jdrosen at cisco.com
http://www.jdrosen.net PHONE: (408) 902-3084
mmusic mailing list
mmusic at ietf.org
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 7338 bytes
Desc: S/MIME Cryptographic Signature
More information about the Standards