- From: Henning Schulzrinne <hgs@cs.columbia.edu>
- Date: Fri, 3 Sep 2010 05:20:12 -0400
- To: "Hermodsson, Klas" <Klas.Hermodsson@sonyericsson.com>
- Cc: Thomas Wrobel <darkflame@gmail.com>, "roBman@mob-labs.com" <roBman@mob-labs.com>, "public-poiwg@w3.org" <public-poiwg@w3.org>
On Sep 3, 2010, at 3:14 AM, Hermodsson, Klas wrote: > > On Sep 2, 2010, at 10:30 , Henning Schulzrinne wrote: > >> Two quick remarks: >> >> - See LoST (RFC 5222) for an example of a global distributed infrastructure for mapping. Such infrastructure is probably well beyond the scope of the W3C. > > Thank you, I will have a look at that. I just threw a very fast glance at it, is it correct in understanding that a likely scenario is to have a well known server to query and which may serve links and services relevant to a certain geographic area? As an example, would that mean that if you are in downtown Tokyo all stores, services, and all other info needs to register themselves to one server/service in order for the user to know where to query? This deserves a longer answer, but the general notion is that there can be any number of server hierarchies, e.g., one tree per service class. There is no "well-known" server as such. There can be several (competing?) server "forests", if that makes sense. In other words, you can slice this both by geography and by service. To be honest, I don't know of a way to avoid having some linkage - you don't want to send a message to a 1000 servers saying "Hi, you got anything about Tokyo?". (If you do, you are asking Bing or Google - this is a (logically) centralized server.) Henning
Received on Friday, 3 September 2010 09:20:46 UTC