Re: [Ecrit] Discussion on draft-winterbottom-ecrit-priv-loc-04

["Rosen, Brian" <[e-mail]>]

> 
>> I am convinced we will need to support an environment where it is not acceptable to have the end device in the path for location.
> 
> We already do support such environments.  In the cellular world, it's more likely that the operator will obtain location (usually with cooperation of the device, but that's besides the point) and routing. The operator might use LoST, but might not, at least in the short to medium term.
Yes, but that only works if the access network = communications network.  The work that James is really hot about includes nomadic VoIP over ISP provided access networks.  

> 
>>  The level of paranoia about that is so high, if we want to make progress in the countries where it exists, we need to do something.
> 
> We already permit proxies to do this.  That's what made the cellular world compliant.
Agree, but see above

> 
>> In some of these environments, it's unacceptable to have the CSP get location.  Using HELD with some non-IP identifier to get location by reference works okay.  What is needed is a way for routing to work.  This can be done in two ways.  One is allowing HELD to return route, which is how this doc does it.  The other is to allow LoST to do LbyR.  I prefer the latter, but would be willing to go along with the HELD mechanism if consensus was to do it that way.  Please don't use the "one less query" argument.  It just moves the route query from the CSP to the LIS.
> 
> Right, it's a query no matter which entity does it, or which protocol is used.
Yeah.  That’s why even though I have a preference for LoST-does-LbyR, I am not really against HELD-returns-route.  

Brian