So, draft-burger-sip-info-02 went over some of the comparisons between
3265 and INFO packages. Perhaps we can head off future confusion by
merging even more 3265-related text from that document into the
info-packages document.
To address Nasir's comments on a more practical level: currently, INFO
is so underspecified that every additional use causes harm to the entire
SIP protocol by introducing new incompatibilities. We can either
deprecate the use of INFO (and be soundly ignored by implementors all
over the world), or we can try to reform INFO to be a (somewhat)
upstanding member of the SIP method pantheon.
And until we do so, we're going to burn unwarranted quantities of time
arguing about the open festering sore that is INFO.
/a
On 10/29/08 12:56 PM, Paul Kyzivat wrote:
I will try to summarize the essence of the arguments:
Establishing the subscription requires more messages to be exchanged.
The subscription then requires state to be maintained, and it must be
refreshed periodically.
The desired uses of info are often things that *might* be exchanged on
any call, but often (usually?) are not. So paying a lot of cost for
the possibility is thought to be a bad deal.
Thanks,
Paul
Nasir Khan wrote:
[Per Keith's suggestion starting a new thread, copied from last one]
Yes Paul, that is what I was trying to get at.
Sorry if all this was already discussed and closed.
Do we have a comaparitive between the two approaches from past?
The reason I brought it out was that there are several similarities
in where we are going with info events and 3265.
I hope we do not undermine the usage of SIP events.
As an example if one comes up with a brand new application that has
some small data to be transferred between UAs and the session is
INVITE based then which approach should one take? Will we have clear
guidelines or leave it open?
IMHO overlapping semantics is also not very good for interoperability.
Is the legacy (mis)use of INFO the most compelling reason for this
approach of bringing it into proper framework?
again apologies for repetition.
regards
nasir
-------------------------------
I'm not certain I understand the point you are making.
Is it that we don't need info events because we could accomplish the
same thing using subscriptions to suitable event packages?
If that is your point, we have been down that path already. Many of
us have taken that position in the past (and may still believe it
today), but there were many arguments that that approach is to heavy
weight. Evidence of that is the continued use of info for DTMF even
though we have an event package for that.
Paul
Nasir Khan wrote:
Maybe it warrants a separate thread but I would like to discuss
section 7.1.3 of this draft where 3265 is talked about as an
alternative.
Since we are talking about the notion of well defined info packages
and the INFO messages
will carry information as a result of an event like an ISUP event or
DTMF event (the name of the draft itself has "event" in it), then the
two User Agents can very well subscribe to these events and use the
notifications for the information exchange.
This will automatically inherit all the good work done around
notification framework and will lay to rest lot of questions and
perhaps save some re-invention.
Regards
Nasir
--
Get the most comprehensive open source SIP application testing
framework at http://sipper.agnity.com <http://sipper.agnity.com/>
------------------------------------------------------------------------
_______________________________________________
Sip mailing list https://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip
_______________________________________________
Sip mailing list https://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip
_______________________________________________
Sip mailing list https://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip
