Re: [core] Review comments on draft-ietf-core-coap-pubsub-05

[Klaus Hartke <[e-mail]>]

Jim Schaad wrote:
> 1.  Welcome to the world of how CoAP defined resolving of URIs.   RFC 6690
> either did not do a good job of explaining things or got them wrong.  There
> is a special case for /.well-known/core in terms of how the resolution of
> links are done.  As I understand things if you do this
>
> GET /ps?rt="temperature"
>
> And get back in the content
>
> </ps/currentTemp>;rt="temperature";ct=50
>
> When you ask for the resource you will look in the following address
>
> GET /ps/ps/currentTemp

Is this really true? I didn't read RFC 6690 for a very long time,
but this would be really surprising. From a quick glance at RFCs
6690/3986/6454 I would assume the following:

* Reference resolution has two inputs: a URI reference to resolve and
a base URI [RFC 3986, Section 5.2].

* Since the above link does not have an anchor parameter, the base URI
is the origin of the target URI [RFC 6690, Section 2.1].

* Assuming the target URI is <coap://example.com/ps?rt="temperature">,
then the origin of the target URI is the triple ("coap",
"example.com", 5683) [RFC 6454, Section 4].

* This presumably means that the base URI is <coap://example.com:5683/> [??].

* Resolving </ps/currentTemp> against <coap://example.com:5683/>
results in <coap://example.com:5683/ps/currentTemp> [RFC 3986, Section
5.2].

Klaus