Dean Willis wrote:
>
> On Apr 20, 2008, at 11:29 PM, Paul Kyzivat wrote:
>
>>
>>
>> Dean Willis wrote:
>>> On Apr 20, 2008, at 3:33 PM, Paul Kyzivat wrote:
>>>>
>>>>
>>>> Dean Willis wrote:
>>>>> On Apr 18, 2008, at 8:27 PM, DRAGE, Keith (Keith) wrote:
>>>>>> 302s recursed by intermediate proxies may be perfectly reasonably in
>>>>>> certain charging environments.
>>>>> Not if they're retargeting. I'd also say that proxy operations
>>>>> that retarget are also generally unreasonable, given that (so far)
>>>>> we have no way to inform the caller of the retargeting.
>>>>
>>>> What is your objection? Is it
>>>> - that the caller may be billed for something he wasn't expecting?
>>>> - that the caller may not want to talk to the new target?
>>> Both, although my bigger concern is the general unanticipated
>>> respondent problem.
>>>> The former is a billing issue, which may or may not be related to
>>>> this discussion, as Keith has pointed out.
>>>>
>>>> The latter can be dealt with via called party identity.
>>> Which, with current specs, only kicks in AFTER the INVITE transaction
>>> has completed.
>>>> In any case it is quite within the normal range of expectation to
>>>> day that you may end up talking to someone other than who you
>>>> thought you were calling.
>>> That's probably true, but it's not a good thing. Why do we keep
>>> trying to reinvent the failings of the pSTN?
>>
>> I am perfectly willing to admit that I might not be imagining how
>> things might work.
>>
>> Perhaps this depends very heavily on exactly what is retargeting and
>> what is not. If I call the Dr and get the answering service is that
>> retargeting? If I call the boss and get the secretary is that
>> retargeting? I I call you on your land line and get forwarded to your
>> mobile line is that retargeting? If I call a help line and get
>> forwarded to a call center worker is that retargeting?
>
> Can they respond with the Dr's credentials? Would their identity service
> put an RFC 4474 in a return UPDATE that matches the To in the original
> request? If so, then it's OK -- the caller is surprised by the response.
> If not, then we have a retargeting. When retargeting occurs without
> infomring the caller, we invariantly have an unanticipated respondent
> scenario.
Well, today if the Dr has sip service with credentials but subscribes to
a PSTN answering service then I presume not. In some other cases
possibly yes.
Even in cases like this one, where it would seem in the abstract that
the new target *ought* to have credentials there will clearly be a
migration issue.
>> In most of these cases, even if my UA has a UI to ask me if I want to
>> continue I will most likely just consider than annoying and
>> reconfigure it to automatically accept all of these retargetings.
>
> even the ones that are spoofed "responses" from an MITM?
Well I would be pleased to have the true spoofs singled out and
highlighted, if it was possible to reliably separate them from the cases
that were clearly intended by the callee. But I don't see that being
possible any time soon.
> Here's the scenario: Alice sends an INVITE to Bob. Charlie sees it, and
> fakes a 200 OK response to Alice before Bob can respond.
>
> Unlikely? Not in a P2PSIP world.
Thats clearlhy a bad case. If it can be distinguished from the others
then great.
However its not entirely clear to me that this is conceptually different
from Bob being offline and having asked Charlie to answer calls for him.
> If Bob forwards his phone to Charlie, Alice needs to see something
> signed by Bob or somebody authoritative for Bob that says the call is
> being redirected to Charlie before accepting a response from Charlie.
I agree this would be a good thing. It doesn't have to be the 3xx response.
>> I can see the value of being able to specify that I want to speak to a
>> certain person and no other, but I think I wojuld only want to use it
>> on rare occasions.
>
> I bet you'll want to use it on every call. You just don't know it yet.
There are just too many cases when I don't know who will be expected to
answer the call. You seem to be assuming that all phones are personal
phones.
>> BTW, when it does matter to me I would prefer to restrict who actually
>> answers the phone, rather than just which phone rings. But I don't see
>> much likelihood of that any time soon.
>
> You have no choice about who answers. What you have a choice about is
> which answer you will accept as valid.
I want the one with the real Dean Willis talking, regardless of what AOR
was used to reach it. (Assuming there *is* a real Dean Willis.)
Paul
_______________________________________________
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
