I agree using HTTP URIs is not good practice in any environment where
people might expect to dereference them. For a purely local system, though,
you would be the only one inconvenienced, and you could always add a web
server to provide that service later, if you felt the need for it.

Otherwise, a different URI scheme is the answer. What I usually use in such
cases is the "tag:" URI scheme: https://tools.ietf.org/html/rfc4151


On 24 May 2018 at 00:31, Laura Morales <[email protected]> wrote:

> I could do that, but can it be considered good practice? Other approaches
> are
>
> - make my own URI scheme, for example <mydocs:name/doc1>
> - make my own URN NID like <urn:mydocs:doc1> but NIDs are supposed to be
> registered
> - the info: scheme has been deprecated, so I should not use
> <info:mydocs/doc1>
>
> I don't think RDF is providing any "best practices" suggestions or
> guidelines for a scenario like this, and it's pretty frustrating because
> not all data need to be dereferenceable and not all data need to have a
> universally unique ID such as ISBN or telephone number...
>
>
>
> Sent: Wednesday, May 23, 2018 at 3:26 PM
> From: ajs6f <[email protected]>
> To: [email protected]
> Subject: Re: Nodes without dereferenceable URIs
> Can you use HTTP URIs that simply don't point to an actual server? (E.g.
> http://lauras.namespace/blah/blah/blah)
>
> If no one tries to dereference them, it's fine if they don't work. If
> someone might try to dereference them, that's when you might have problems.
>


-- 
Conal Tuohy
http://conaltuohy.com/
@conal_tuohy
+61-466-324297

Reply via email to