[Jingle] Typo and possible clarification in XEP 0166
stpeter at stpeter.im
Mon Apr 20 10:40:14 CDT 2009
-----BEGIN PGP SIGNED MESSAGE-----
On 4/20/09 8:29 AM, Peter Saint-Andre wrote:
> On 4/19/09 10:36 PM, Justin Uberti wrote:
>> One thing I noticed is that we don't have any defined semantics for
>> handling ties on the call initiation (i.e. if A and B call each other at
>> the same time). This happens more often than you would expect, and there
>> are some easy things we could do to handle this scenario (e.g. lowest
>> JID or session id wins). To have this actually work across all clients,
>> we would need to explicitly specify this. Thoughts?
> Good point!
> Given the possible problems with JID comparison (not everyone does
> stringprep correctly), I think that comparing session IDs might be
> safer. But that means we might want to specify the allowable characters
> for session IDs a bit more carefully (e.g., say that they need to match
> the NMTOKEN datatype).
> ... Well, checking the schema I see that we already say this:
> <xs:attribute name='sid' type='xs:NMTOKEN' use='required'/>
> So that would work.
I've adjusted the text as follows...
7.2.16 Tie Breaking Related to Jingle Actions
It is possible that the same Jingle action can be sent at the same time
by two different parties. There are two possible scenarios:
No existing session
If there is no existing session and both parties simultaneously send
a Jingle session-initiate message, the action with the lower of the two
session IDs MUST overrule the other action, where by "lower" is meant
the session ID that is sorted first using "i;octet" collation as
specified in Section 9.3 of RFC 4790 . The party that receives the
session-initiate action with the higher of the two session IDs MUST
acknowledge the action or return an error condition that would normally
be returned when receiving a session-initiate message, and the party
that receives the session-initiate action with the lower of the two
session IDs MUST return a <conflict/> error to the other party, which
SHOULD include a Jingle-specific condition of <tie-break/>.
In the context of an existing session, the action sent by the
initiator MUST overrule the action sent by the responder; i.e., both
parties MUST accept the action sent by the initiator and the initiator
MUST return a <conflict/> error to the responder for the duplicate
action, which SHOULD include a Jingle-specific condition of <tie-break/>.
In both scenarios, the error to be returned is <conflict/>, as shown in
the following example.
Example 34. Initiator returns conflict error on tie-break
<iq from='romeo at montague.lit/orchard'
to='juliet at capulet.lit/balcony'
Does that seem acceptable?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the Jingle