- From: Andy Seaborne <andy@apache.org>
- Date: Fri, 2 Apr 2021 14:39:05 +0100
- To: public-rdf-comments@w3.org
I agree with Pierre-Antoine.
The wording could be tightened up.
RDF Concepts says:
"IRIs in the RDF abstract syntax MUST be absolute, and MAY
contain a fragment identifier."
but (RFC 3987) the absolute-IRI rule doesn't have a fragment so at least
the wording should be "may also contain" but "conform to the "IRI"
syntax is enough here.
While in the area:
By RFC 3986/87:
http:abc
is an IRI (and absolute-IRI).
It is the HTTP URI scheme (RFC7230 sec 2.7.1) that requires the
authority to get http://host/path.
There needs to be an additional requirement to follow the rules of the
specific scheme.
"and SHOULD conform to any additional rules of the URI scheme, and URN
namespace if a URN."
("SHOULD" because this is a change).
For "http:", syntax parsers can be recommended to resolve same-scheme
URIs in non-strict mode (RFC3986/sec 5.2.2) and then a reference written
"http:abc" and well-formed base http: URI will produce an HTTP URI.
URNs have additional rules and so do the URN namespaces (e.g. UUID URNs)
Is it worth defining a piece of terminology e.g. "RDF Reference" to
collect the points together?
Andy
On 31/03/2021 17:48, Pierre-Antoine Champin wrote:
> Hi all,
>
> RDF specifications talk a lot about IRIs. Sometimes, the distinction is
> made between absolute IRIs and relative IRIs. I discovered some time ago
> that this is inaccurate.
>
> According RFC3987 [1], IRIs can not be relative, only *IRI references*
> can. Therefore, "absolute IRI" is redundant (but correct), while
> "relative IRI" should be "relative IRI reference".
>
> Note that, from what I saw, this impacts the following documents:
>
> RDF 1.1 Concepts and Abstract Syntax
> RDF 1.1 Turtle
> RDF 1.1 TriG – RDF Dataset Language
> RDF 1.1 XML Syntax
>
> the others make use of "IRI", "absolute" and "relative" in a way that's
> consistent with RFC3987.
>
> pa
>
> [1] https://www.ietf.org/rfc/rfc3987.txt
>
Received on Friday, 2 April 2021 13:39:19 UTC