Re: [T2TRG] draft-hartke-t2trg-ciri-00 review

[Klaus Hartke <[e-mail]>]

Dave Thaler wrote:
> Once you have COAP/HTTP proxies, you pull in all that baggage.

Right. Currently, converting a CIRI to CoAP options is defined as:

1. Recompose the CBOR-encoded IRI components to an IRI
2. Convert the IRI to a URI according to RFC 3987
3. Decompose the URI to CoAP options according to RFC 7252

It sounds like a good idea to rewrite section 4.2 to output a URI and
define the conversion as:

1. Recompose the CBOR-encoded IRI components to a URI
2. Decompose the URI to CoAP options according to RFC 7252

> Furthermore, the conventional wisdom is that there's no problem that
> putting IRIs on the wire solves that isn't better solved by putting URIs
> on the wire.

The Semantic Web relies extensively on IRIs, and some formats like
Turtle put IRIs "on the wire" rather than URIs. E.g.:

   @prefix loc: <http://example.org/location/> .

   loc:Montréal a loc:city .

Would this be considered "UI layer"? If not, I'm not sure if URIs
would improve things:

   loc:Montr%C3%A9al a loc:city .

(Is that even valid?)

Since the CoRAL textual format [1] uses IRIs in a similar way, some
advice on IRIs in formats intended to be read and written by humans
would be very welcome.

Klaus

[1] https://tools.ietf.org/html/draft-hartke-t2trg-coral-06