[JDEV] Re: [oob-jig] voice chat and video chat

[ryutaroh at it.ss.titech.ac.jp]

Thanks for your comments.

From: Jacek Konieczny <jajcus at bnet.pl>
> Please, anything but H.323. It is the most horrible protocol I know.

I don't adhere to H.323. I want a standard method starting voice chat
and video chat within the Jabber protocol.

I don't want to force Jabber clients to include the H.323 protocol
stack, either.

> I would rather see the voice and video connections innitated through
> jabber as SDP (session description protocol) message sent in a special
> namespace object. With SDP it would be easy to cooperate with SIP
> protocol, and all the signaling could be done via Jabber.
> Using just RTP and SDP (or maybe SDP is not needed, tha same information
> may be passed in special XML namespace) we could use PASS, I don't think
> it would be possible with H.323.
> 
> And H.323 implementation in Jabber clients would be very hard. Starting
> RTP-talking software from client using data from SDP description (or
> some XML description) would be much easier.
> 
> We can also use jabber:iq:oob and SIP protocol, but IMHO Jabber with its
> XML stream may be much better for signaling, and RTP is used for
> transmision in any signalling protocol anyway.

My consideration was aimed to reduce the changes in protocol and the
effort of implementation by client developers. To achieve this goal, a
client simply runs a (SIP or H.323) VoIP application upon receiving
"iq" with non-http URI in the jabber:iq:oob namespace, and a Jabber
client does not have to do signaling and voice&video coding by itself.

There are SIP or H.323 applications on almost every platform, at least
on Windows, Mac, and open source UNIXs.

Do you prefer signaling done within the Jabber protocol to letting
an outside application do signaling?  Signaling in the Jabber protocol
needs protocol addition and effort in client implementation. I am
afraid that voice&video chat support will never be realized within
Jabber.


Best regards,

Ryutaroh