Re: GEOPRIV and the W3C Geolocation API

Hi Doug,

>> I'll put my motives up front: I want the W3C Geolocation API to be 
>> compatible with location specifications that GEOPRIV has developed and 
>> is working on.
> I have no problem with that as so long as it benefits the web, solves a 
> problem we have, and is as simple and elegant as possible.

I certainly agree with those objectives.  Part of what I'm trying to 
argue is that there's appreciable benefit to be had without really any 
additional complexity.

>> To help demonstrate what this integration might look like, I've 
>> extended Doug's geolocation extension for Firefox so that it can use 
>> location it gets via the IETF HELD protocol:
> Is there a HELD server that I can test this against?

Sure.  The extension is pre-configured to use this LIS:
<http://geopriv.dreamhosters.com/cgi-bin/lis.pl>
(Full disclosure: I also wrote the LIS, and it's got some bugs.  But it 
has been tested with other clients.) It uses some rough IP geo 
techniques to deliver a guess at a lat/long, as well as a civic location.

It's worth noting that in this case, because of the geo techniques being 
used, the civic location is significantly more accurate than geodetic.

> BTW, if this sort of UI is required to make HELD useful, i think we have 
> a false start.

Of course, that level of UI isn't *necessary*.  This being a 
techi-focused extension, it's intended to illustrate the flexibility of 
preferences that users can communicate to location-based web pages.

> We punted on reverse geolocation as there is a cost-to-implement as 
> well.  The mailing list has a thread on why.

Of course, reverse geocoding has lots of costs.  That's not what we're 
talking about.

All I'm suggesting is that the API be able to carry civic location. 
This location can be derived from geo, but there are also positioning 
mechanisms that can produce civic addresses natively (e.g., wire-map 
databases for fixed networks, wireless base-station based location for 
wireless networks).

Enabling the API to carry civic location is *not* the same as requiring 
reverse geocoding.


>> I admit that GEOPRIV location objects have a lot of expressive power, 
>> and not all of that is relevant to web applications.  However, if 
>> we're willing to step a little bit beyond our pre-conceptions about 
>> lat/long location and yes/no privacy, then we can get a lot of benefit 
>> out of GEOPRIV at very little cost.
> 
> Please give a use case.

For civic, Martin has given several examples, in the unfortunately-named 
"reverse geocoding" thread.  For instance, a page could help fill out 
form fields.

For rules, Alissa and John have also given several examples, in the 
various privacy threads.  Think of it as a partial formalization of the 
privacy agreement process, focused on the user expressing his 
preferences instead of just the site laying down its rules.

Cheers,
--Richard

Received on Tuesday, 11 November 2008 02:23:36 UTC