RE: VS: VS: [Sipping] FW: I-DACTION:draft-sawada-sipping-sip-offeranswer-01.txt (Comments 4 to 6)

["Christer Holmberg \(JO/LMF\)" <christer.holmberg at ericsson.com>]

Hi,

I had a look at chapter 10.2 of RFC 3725, and I think I found an example to explain my issue.

Look at message 14, where "Called party" sends an offer in the 200 OK response. This SDP is then forwarded by the controller as an offer towards "Pre-paid user".

Now, the previous active local SDP of "Called party" (sent as an answer to "Controller" in message 2) contained the negotiated codecs between "Controller" and "Called party". If "Called party" would simply send that SDP as an offer in the 200 OK it is not sure that the session with "Pre-paid user" would succeed, because it may not support the codec set which was used before. So, in this case I guess it is assumed that "Called party" will send an offer as for a new call, including all codecs it supports.

Now, my question: since "Called party" may have no clue about the scenario, and no clue about all the other entities involved, how does it know that it "wants" to send such offer at this point?

Regards,

Christer



 

-----Original Message-----
From: Paul Kyzivat [mailto:pkyzivat at cisco.com] 
Sent: 27. kesäkuuta 2006 16:46
To: Christer Holmberg (JO/LMF)
Cc: Takuya Sawada; sipping at ietf.org
Subject: Re: VS: VS: [Sipping] FW: I-DACTION:draft-sawada-sipping-sip-offeranswer-01.txt (Comments 4 to 6)



Christer Holmberg (JO/LMF) wrote:
> Hi,
> 
>>>> Precisely which scenarios do you mean. It is pretty crazy to expect 
>>>> such behavior, but I am not too surprised if there are scenarios 
>>>> like
> 
>>>> that.
>>>> I think there are a cases that require a UAS to be psychic. (Those 
>>>> related to REFER often feel that way. Dale has recently written 
>>>> something on that.)
>>> [CHH] It's quite a while since we discussed this, but hopefully 
>>> those who pushed for the specfic behavior can correct me if I'm 
>>> wrong. The scenario was something like the following:
>>>
>> User A receives an hold re-INVITE. Later (when some transfer 
>> functions
>>> have taken place) user A receives an empty re-INVITE. Now, in order 
>>> to
> 
>>> make things work, user A is supposed to un-hold the session and send 
>>> an SDP offer as for a new session - not simply return the SDP from 
>>> the
> 
>>> previous offer/answer transaction.
>> This doesn't require User A to be psychic - only to keep stating its
> desires at each opportunity.
>> In this scenario I believe User A was *put on hold* - he didn't 
>> *want*
> to be on hold. So at each opportunity (like the offerless reinvite) it 
> simply
>> states what it wants - which is *not* to be on hold.
>>
>> If, after A had been put on hold it decided to initiate its own hold
> action (perhaps because another call was received), then when the 
> reinvite is
>> received it might indeed specify SENDONLY. That wouldn't be what the
> sender of the reinvite wants to hear, but it is the current situation.
> (Of course
>> when A decided to do it own hold it probably should have done its own
> reinvite specifying INACTIVE.)
> 
> Ok, so what is the "desire" of an MGC (or any other entity without 
> direct user control)? How does an MGC know whether he "wants" to be on 
> hold or not?



It needs to figure it out based on interactions with the other side.

For a "regular" UA, I think a reasonable way to make the determination is based on whether the media stream(s) are connected to a primary local source, sink, or both, or neither. I think the same works for an MGC.

Signaling on the non-sip side might influence this state.

	Paul

_______________________________________________
Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use sip-implementors at cs.columbia.edu for questions on current sip
Use sip at ietf.org for new developments of core SIP