RE: ID48 - Relationship of profile to validation

Hi Ruben, all,

On Wednesday, August 16, 2017 8:07 PM, Ruben Verborgh [mailto:Ruben.Verborgh@UGent.be] wrote:

> > All a profile needs formally is an identifier - with the semantics of comparability.
> 
> I agree, but may I suggest a slightly more specific definition for our purposes?
> 
> Given the context we operate in, it might be good to agree that
> a profile needs an IRI as identifier (with comparability semantics),
> and, whenever possible, this IRI should be an HTTP URL.
> 
> That way, the identifier of a profile can be dereferenced,
> which allows a server to further detail that profile,
> for instance, through SHACL, HTML,
> or even a standard that is yet to be created.

I see two possible ways here:
1) We have one identifier for the profile document _and_ the different schemas that implement that profile and use content negotiation to serve the most appropriate version.
2) we have one identifier for the profile document and one identifier for each schema implementing the profile and link from the profile document to the schemas using the Link-header and (in html) <link>-elements.

Both ways have their pros and cons. I'd tend to suggest 2) since the profile document and the associated schemas might not really be considered the same thing and thus should have different URIs. What do you think?

Best,

Lars

Received on Thursday, 17 August 2017 20:00:46 UTC