[Jingle] do we need Raw UDP?

Arc Riley arcriley at gmail.com
Thu May 21 13:32:22 CDT 2009


ICE can also be overkill in some non-SIP use cases, such as with link-local
XMPP where it's known that the other party can connect directly.


On Thu, May 21, 2009 at 6:21 AM, Paul Witty <paulrw at codian.com> wrote:

> Olivier Crête wrote:
>
>> On Mon, 2009-05-18 at 12:59 -0600, Peter Saint-Andre wrote:
>>
>>
>>> The subject of this message says it all: do we really need Raw UDP, or
>>> is XEP-0177 useful only for bootstrapping and unit testing? Given that
>>> the Raw UDP candidate will be included in ICE negotiation anyway, I
>>> don't see much of a need for XEP-0177 at this point, so I'm wondering if
>>> perhaps we retract it and proceed only with XEP-0176 (ICE-UDP) as the
>>> preferred datagram transport.
>>>
>>>
>>
>> I guess one argument for rawudp is integrating XMPP support into legacy
>> sip gateways without having to implement Full ICE there (since it XMPP
>> forbids ICE-lite).
>>
>>
>>
> We gateway Jingle <-> SIP by terminating the media at the gateway, which
> allows us to perform ICE on the Jingle side even when the SIP side does not
> support it.  However, the design of Jingle to be SDP means that we don't
> need to do this, and it would be nice to avoid doing so where possible, but
> this requires Raw UDP support for almost all existing SIP implementations.
>
> In my opinion, we should keep both, with wording in XEP-0177 that clients
> which support Raw UDP SHOULD support ICE UDP.  Even if it is only useful for
> bootstrapping, unit testing and talking to gateways, it's still useful.
>
> --
>
> Paul
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.jabber.org/pipermail/jingle/attachments/20090521/e48c3fbf/attachment.htm>


More information about the Jingle mailing list