Re: [core] draft-ietf-core-link-format-11

On Apr 21, 2012, at 6:25 AM, Klaus Hartke wrote:

> Hi Zach,
> 
> On 15 April 2012 19:47, Zach Shelby <zach@sensinode.com> wrote:
>> What this means in practice is that the target for this relation has to be on the same origin server. The result is that if the target is an absolute URI, then the host part should be interpreted as a virtual host name (and thus should be included in the Uri-Host option). [...] There is no reason your virtual host name couldn't be really short though and thus just as efficient as using the Uri-Host field.
> 
> Do you mean a server like "sensor1.example.com" could advertise a
> virtual host "xy" rather than "temp.sensor1.example.com" in its
> well-known core?

There is no "sensor1.example.com" in the first place as you are not using DNS in this case. The origin server in this example would be identified by [2001:DB8::abcd]:5683. If /.well-known/core returns a link  <coap://xy/res> then maybe a cache should identify that as coap://xy.[2001:DB8::abcd]/res or in some other way associated with the IP address:port as this is the only unique identifier it could possibly have for the server. I do agree that is a bit awkward, other ideas?

Zach

> 
> What happens when two different servers both advertise a virtual host
> "xy"? Can a client with a cache use a stored response that is the
> result of retrieving resource "/res" of virtual host "xy" at
> [2001:DB8::abcd] to satisfy a request for resource "/res" of virtual
> host "xy" at [2001:DB8::9999]? The request URI is <coap://xy/res> in
> both cases.
> 
> 
> Klaus
> _______________________________________________
> core mailing list
> core@ietf.org
> https://www.ietf.org/mailman/listinfo/core

-- 
Zach Shelby, Chief Nerd, Sensinode Ltd.
http://www.sensinode.com
http://zachshelby.org  - My blog "On the Internet of Things"
http://6lowpan.net - My book "6LoWPAN: The Wireless Embedded Internet"
Mobile: +358 40 7796297

Re: [core] draft-ietf-core-link-format-11 - discover virtual hosted resources