[foaf-protocols] WebID spec: supported RDF formats (bis)
Bruno.Harbulot at manchester.ac.uk
Sun Jul 18 19:16:07 CEST 2010
On 18/07/2010 18:07, Nathan wrote:
> Bruno Harbulot wrote:
>> On 18/07/2010 17:19, Seth Russell wrote:
>>> I would mandate that the service providing the WebID and making the
>>> remote profile document support *all* the reasonable formats. To those
>>> i would optionally add suport for JSON as that is probably going to be,
>>> me hopes, the best in the long run. Then the verifying agent need
>>> only support which format they are comfortable with.
>> I think it's much easier to do the other way around. There will be
>> fewer libraries for verification agents, and that's where the
>> processing takes place.
> What we're dealing with here is the web of Linked Data, and on that we
> can't mandate that one serialization should be used over another, or if
> you provide one serialization you must also provide several others.
> The only place any standardization can happen is on the Linked Data
> Client side of things, stating that a Linked Data Client *must* support
> all the known serializations; doing it any other way makes the web of
> Linked Data not a web anymore and is impossible to deploy, it's already
> more than enough to expect somebody to understand one serialization of
> RDF in order to publish Linked Data without all the extra constraints
> this would impose; whereas it's entirely practical to expect a
> programmer who understands RDF to add support for several serializations
> in order to conform and make this thing work.
(I'm not sure whether you agree or disagree with what I said.)
I'm just saying the Verification Agent MUST support *some* known
serializations (specifically RDF/XML and HTML+RDFa) and MAY support
anything else on top of that. Note that, in this case, the Verification
Agent *is* the client.
I'm also saying we need to mandate at least one of these formats from
the WebID provider side (which will be the server). The reason for this
requirement is because we're dealing with a client that's an automated
agent (the verification agent).
While with a browser user-agent, I can click on 'save as/open with' when
I get a content-type the browser doesn't understand, the automated agent
doesn't has to know how to make things work (otherwise, the verification
will simply fail).
More information about the foaf-protocols