Re: [RFC 4235 & draft-kuthan-sip-derive] 481 "Call doesn't exist" for initial SUBSCRIBE ?

Alan Johnston wrote:
> Hold on a minute - are we sure the notifier should send a 481?
> 
> I have always thought about the call-id, and tag Event package 
> parameters as essentially a filter - that is only send me notifies about 
> this particular dialog.  If I subscribe to an event package with a 
> filter, but nothing matches the filter at that moment, would we normally 
> expect the subscription to be refused?  Instead, I would expect an 
> immediate notification saying subscription is created but then not get 
> any notifications until the filter conditions are met.   Or is my 
> understanding of how filtering works wrong?

I'm inclined to think of it in much this way, but there are some odd 
twists to it:

- These options (or filter) also serve as data used for authentication
   and authorization. Namely, possessing a valid dialog id is considered
   a form of identity, that is then coupled to authorization to subscribe
   with a limitation on seeing only data about that dialog.

- IMO it can be assumed that if someone supplies the notifier with what
   should be one of its own dialog ids, and it isn't currently defined,
   then it won't be valid in the future, and so should not serve as a
   valid identification. So such a subscription should fail, except
   perhaps if the subscriber can be authorized to subscribe on some
   other basis.

On that basis, I can see a 403 response if the dialog id isn't valid and 
the subscriber has no intrinsic authorization to subscribe.

I understand the reasoning behind 481 but I am a *bit* troubled by it. 
It is a distinct semantic for 481, though one that is unambiguous. But 
481 already has three distinct meanings, so adding another doesn't seem 
like a great idea.

	Thanks,
	Paul

> As such, I think it still works - if the subscriber does not receive any 
> notifications that match the dialog, he does not get the DERIVE 
> authentication he is after.
> 
> Thanks,
> Alan
> 
> Rohan Mahy wrote:
>> Be generous in what you expect, but send a 481.   ;-)
>> thanks,
>> -rohan
>>
>> On Nov 12, 2008, at 2:31 PM, Adam Roach wrote:
>>> We can argue about whether a UA should send a 481 in this case. What 
>>> we can't do is change what any given deployed UA actually does in 
>>> this case.
>>>
>>> The difference between "should" and "does" is critical when trying to 
>>> leverage deployed behavior.
>>>
>>> /a
>>>
>>> On 11/9/08 4:02 PM, Rohan Mahy wrote:
>>>> Hi Jiri,
>>>>
>>>> I am just saying that you if you send a 481 in this case, the 
>>>> semantics are clear and consistent with some other event package 
>>>> that uses call-id attributes. If you receive a 481 in this case you 
>>>> can act accordingly. You could also get back a NOTIFY with a body 
>>>> with full state with no dialogs, but this is less useful.
>>>>
>>>> thanks,
>>>> -rohan
>>>>
>>>> On Nov 9, 2008, at 1:45 PM, Jiri Kuthan wrote:
>>>>
>>>>> Hi Rohan,
>>>>>
>>>>> is this our wishful thinking (I would like it too for it simplifies 
>>>>> few
>>>>> things) only, or is there a normative reference (which I can't find).
>>>>> The way I read the related stuff is I get a 200 for the SUB back, and
>>>>> an "empty" NOTIFY.
>>>>>
>>>>> -jiri
>>>>>
>>>>> Rohan Mahy wrote:
>>>>>> Hi,
>>>>>> The UA should respond with a 481.  In the case of an out of dialog 
>>>>>> request the *meaning* of the response in this context is easy to 
>>>>>> understand. A receiving UA realizes that it referred to some 
>>>>>> dialog that does not exist. There are far too few response codes 
>>>>>> for us to create a new one in this situation, and it is unlikely 
>>>>>> that an automaton will be able to recover from this error if it 
>>>>>> has a different code.
>>>>>> thanks,
>>>>>> -rohan
>>>>>> On Oct 25, 2008, at 10:26 AM, Iñaki Baz Castillo wrote:
>>>>>>> This SUBSCRIBE arrives to Alice's UA which is not aware of that 
>>>>>>> dialog. Which
>>>>>>> response should it reply? The flow suggests "481 Call/Transaction 
>>>>>>> doesn't
>>>>>>> exist", but... is it correct?
>>>>>>>
>>>>>>> AFAIK, a 481 should be replied when an *in-dialog* request 
>>>>>>> arrives to an UAS
>>>>>>> which is not aware of that dialog. But in the above case we have 
>>>>>>> an *initial*
>>>>>>> request and the 481 refers to the specific dialog in the "Event" 
>>>>>>> header.
>>>>>> _______________________________________________
>>>>>> 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
>>
>>
> 
> _______________________________________________
> 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