[Oslc-pm] monitoring agent availability status

[John Arwe]

> When I look at the definition of the state property, it says:
>   <rdfs:range rdf:resource="
http://open-services.net/ns/auto#AutomationRequest
> " />
>   <rdfs:range rdf:resource="
http://open-services.net/ns/auto#AutomationResult
> " />
> 
> But on the specification page it says:


You have found a bug in the Automation vocabulary document; I will queue 
it up to that mailing list.  Most people have concentrated on the spec, so 
this is unsurprising yet inconvenient for you/me.
IMO neither range should be there.  I cannot be sure which URL you were 
looking at since you didn't specify it, but since rdfs:range is not in the 
HTML vocabulary document that I get [2] when dereferencing the state URI 
(I just clicked on its value [3] in your email and adjusted the fragment 
to be state [4], and the server redirects) I assume you found the RDFS 
itself [5] from which the HTML is generated via XSLT.  The HTML contains 
no equivalent to the rdfs:range information, you will note.  That is 
intentional IMO.
If you stick with the generated HTML description you should see the 
intent, and that should be consistent with the spec.  The portion "It is 
expected that this will be a resource reference to a definition of a valid 
automation request state on the service provider." is a really tortured 
way of saying that we expect whatever URL is used to be a normal 
vocabulary term (an rdfs:Class , in particular) like the set defined in 
the spec.... meaning that (regardless of the URL) you can successfully 
HTTP GET some representation of it.... and that the URL can be any state 
the implementation considers valid.  I've also queued that tangled 
sentence for editorial attention by Automation.

rdfs:Range is used in a way other than what you appear to think.  Details 
at [1]
Most people read "RDF Schema" and assume it works like "every other" 
schema language they've encountered, e.g. like XML Schema, insofar as the 
schema language specifies machine-readable constraints on "valid" 
instances, and (usually) a way to assess the validity of an instance, 
where "validity" is defined in terms of adherence to a set of constraints. 
 This (mis-)interpretation is reinforced by the resource definition table 
column label (Range) which also has a different usage (to set 
-expectations- for implementations, not for inferencing).
More succinctly, 
1: validity = f( instance , constraints) 
2: schema language = syntax for specifying constraints

This is NOT how RDFS works.  RDFS is used for inferencing.  The schema 
syntax (rdfs:range and friends) is used to specify rules by which new 
triples may be inferred from existing triples.  I.e.:
1: inferrable triples = f( input triples , RDFS triples )
or 2: g' = f( g , RDFS triples )  where g is an RDF graph and g' is the 
resulting graph containing the union of g and the newly inferred triples.
In particular, RDFS defines no notion of validity.  OWL can be used, with 
work, for specifying constraints, but that is not its main job either. 
IIRC OWL has a notion of "consistent with" that loosely corresponds to our 
traditional notion of validity.

[1] http://www.w3.org/TR/2004/REC-rdf-schema-20040210/#ch_range
[2] http://open-services.net/ns/auto/auto.html#state
[3] http://open-services.net/ns/auto#new
[4] http://open-services.net/ns/auto#state
[5] http://open-services.net/pub/Main/AutomationVocabulary/auto.rdf

Best Regards, John

Voice US 845-435-9470  BluePages
Tivoli OSLC Lead - Show me the Scenario

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://open-services.net/pipermail/oslc-pm_open-services.net/attachments/20120831/11fc062e/attachment-0003.html>