Re: [Geopriv] Re: [Sip] teasing apart: http as a GEOPRIV using protocol

From: Hannes Tschofenig ^lt;>
Date: Thu Jul 27 2006 - 15:27:23 EDT

Hi Henning,

I would very much appreciate to treat this topic in the way you proposed
it below.

A few minor comments:

Henning Schulzrinne wrote:
> With hindsight, I think the term 'using protocol' has been the source of
> more word-lawyering than engineering insight. I personally find the
> distinctions
> - first person vs. third person (i.e., find out location about myself
> vs. find out location about some third party)
> - location vs. location + identity
> - token-based authorization vs. identity-based authorization
> more helpful.
> Token-based authorization hands the recipient a key that allows him to
> get a location object, without caring as to who is retrieving the
> object. ("Anybody seeing this SIP request can convert the token into an
> LO.") Identity-based naturally uses some authenticated identity to
> release that information.

In past common policy discussions we also differentiated between
limited use addresses and trait-based authorization. A token would match
the latter.

In some sense we seem to be using limited use addresses in most of the
cases when we talk about the location-by-references. The difference
between limited use addresses and trait-based authorization is small in
this particular usage environment.

  Strictly speaking, this is only a question of
> a level of indirection, since proving identity usually involves proving
> possession of a secret.

I guess there are some further differences with regard to the lifetime
of the different identifiers, the ability to revoke rights, the number
of entities that one can hide behind these identifiers and the storage

> We've been trying to get at some of these distinctions with the sighting
> vs. using protocol terminology, but that distinction also seems to
> degenerate into discussions more suited for a law school or theology
> department than an engineering organization.
> For example, by strict definition, LoST would be a 'Using Protocol'
> since it "carries a Location Object", but I hope we all agree that this
> is silly. We had earlier agreed that the location-by-DHCP also were not
> Using Protocol, even though they obviously carry a location object of
> some form.

Currently LoST is not a using protocol since it does not carry a PIDF-LO
although some folks would like to use a PIDF-LO; HELD was a using
protocol before we turned the problem into a Location Configuration
problem; RELO never was a using protocol;

Isn't this confusing?

> Privacy issues apply iff
> - third person
> - location + identity (not just location)
> The L7 discussion was about 'first person' and 'location only' (although
> the latter is subject to debate, depending on whether one considers an
> IP address of identifying an individual or not).

That was before we added the Location-by-Reference and subscription-URI
concept ....

> Depending on the circumstances, SIP-based retrieval can either involve
> privacy concerns (subscribe to or not (subscribe to
> location of one-time random unlinkable identifier
> There can also be different privacy concerns
> depending on the perspective. For example, if a SIP request contains
> both an identity and the location retrieval reference, this raises
> privacy concerns for the proxy, but the LIS may not have any such
> concerns since it may only be the repository of these randomly-named
> objects, without tie to an identity.


> Henning
> _______________________________________________
> Geopriv mailing list

Geopriv mailing list
Received on Thu, 27 Jul 2006 21:27:23 +0200

This archive was generated by hypermail 2.1.8 : Thu Jul 27 2006 - 15:34:26 EDT