[ckan-dev] ckanext-rdf: Consolitdated RDF handling for CKAN

[James Gardner]

Hi Adria,

> * [2011-06-01 16:02:20 +0100] Adrià Mercader<amercadero at gmail.com>  écrit:
>
> ] Hi all,
> ] Related to this topic, which do you think would be the more "natural"
> ] routing to a package RDF representation?
> ]
> ]  * /package/{id}.rdf
> ]  * /api/rest/package/{id}.rdf
> ]  * /record/{id}.rdf    (semantic.ckan.net uses this one)
> ]
> ] I'd say I prefer the first one, but maybe is not conceptually correct?
>
> The semantic.ckan.net choice is deliberate but doesn't follow the ckan
> data model (ckans are just one possible source of metadata). The
> reasoning is that the document describes a record which is about a
> dataset/package.
>
> In a ckan-specific scenario I would agree with you the first is
> better, but keep in mind that the thing being described is going to be
> /package/{id} and it will have various representations, .html, .rdf,
> .ttl, .json, etc.

I agree. As part of the model refactor we'll probably need to change 
this but for now let's use the first option. Let's not worry about 
/package/{name}.rdf at all, I'd prefer not to support an alias, we can 
just link via the ID.

> Also keep in mind that this arrangement (which is configurable) does
> not follow the cabinet office guidelines, which say you should use
> /id/package/{id} for the identifiers and
> /doc/package/{id}.(rdf|html|whatever) for the representations...

Handy to know, thanks Will.

James