Re: [dispatch] draft-winterbottom-dispatch-locparam

On 7/18/15 9:49 PM, James Winterbottom wrote:
> Hi Paul,
>
> Thanks for the comments.
>
> I will get to the questions in a second. I just wanted to say that draft is kind of a stawman to indicate what we would like to do and what we had in mind, we are totally aware that it needs work, we just want to get it to the right group to get the work done.

OK. It helps to know that.

> I can answer the second set of questions quickly, I will need to look up the 3325 question.
>
> Yes, somebody could lie, but if we were to assume some kind of trusted network then there are a multitude of other things that they could also lie about couldn’t they?

Ideally there would be some way of signing the assertion, proving a 
right to use that name. Something like the stuff STIR is working on. But 
the STIR work won't cover this.

> Yes, I was referring to the host from 3261, and I agree that an IP address make little sense. Is there are a better way to specify what we are looking for?

If IP makes no sense, then you could simply require 'hostname'.

> I was thinking some kind of whitelist if terms of preference (how this gets created I see as being largely out of scope), I am not sure how much sense it makes to have a blacklist.

If that is the case, then I guess maybe a proxy inserting this may not 
want to insert the actual hostname of itself - it may want to insert a 
more general hostname for the domain it represents, since otherwise the 
whitelist might get long, or else might have to consist of patterns that 
match multiple hostnames.

	Thanks,
	Paul

> Cheers
> James
>
>> On 17 Jul 2015, at 12:51 am, Paul Kyzivat <pkyzivat@alum.mit.edu> wrote:
>>
>> James,
>>
>> I was just looking at this draft. I don't understand what assumptions it is making regarding trust model. Is this intended specifically for 3gpp and its trust model? Section 4 says:
>>
>>    If a proxy receives a message
>>    from an untrusted source with the loc-src parameter set then it MUST
>>    remove the loc-src parameter before passing the message into a
>>    trusted network.
>>
>> Are we assuming a Spec(t) as defined in 3325?
>>
>> And is that enough? The obvious issue is the potential for somebody to lie about the location source.
>>
>> Also, the location is specified as a 'host'. I assume you mean that as defined in 3261, where it is defined as:
>>
>>   host = hostname / IPv4address / IPv6reference
>>
>> Is it really meaningful to use an IP?
>>
>> Is the expectation that the recipient will simply use these with a whitelist or blacklist to determine preferences? Or else how will it be used?
>>
>> 	Thanks,
>> 	Paul
>>
>> On 6/23/15 5:11 AM, James Winterbottom wrote:
>>> Hi All,
>>>
>>> I a notice about this draft a few weeks ago but haven’t seen and follow up discussion.
>>> This work is required in ETSI and I believe that as I understand from some 3GPP folks they would like something similar to this also.
>>>
>>> https://tools.ietf.org/html/draft-winterbottom-dispatch-locparam-00
>>>
>>> I am just not sure where the right final home for it is.
>>>
>>> Cheers
>>> James
>>>
>>> _______________________________________________
>>> dispatch mailing list
>>> dispatch@ietf.org
>>> https://www.ietf.org/mailman/listinfo/dispatch
>>>
>>
>> _______________________________________________
>> dispatch mailing list
>> dispatch@ietf.org
>> https://www.ietf.org/mailman/listinfo/dispatch
>
>