RE: Define/change the range of "supportedProperties" (ISSUE-37)

On Monday, March 10, 2014 9:01 PM, Ruben Verborgh wrote:
> >>>> _:class hydra:supportedProperty foaf:name;
> >>>>     hydra:propertyRestriction [ hydra:property foaf:name;
> >>>>                           hydra:required true ].
> >>
> >> That's great. I like that a lot.
> >> Then the range of supportedProperty is simply rdf:Property.
> >
> > Would you equally like it if the first triple wouldn't be there?
> 
> Almost; "propertyRestriction" maybe might not be the correct term then.

No, it wouldn't.


> But how would this continue?
> Would it be
> 
> _:class hydra:propertyRestriction
>     [ hydra:property foaf:name; hydra:required true ];
>     [ hydra:property foaf:name; hydra:writeonly true ].
> 
> i.e., independent restrictions like in OWL, or
> 
> _:class hydra:propertyRestriction
>     [ hydra:property foaf:name; hydra:required true, hydra:writeonly
> true ].

Hmm.. I would say the latter but haven't really thought about the former. Is
there an advantage of using the former?


> i.e., a property definition.
> Because in the latter case, "restriction" is probably not the right
> name;
> and then we're back at the start; because it is in fact a proxy.

Mmhhmm


> > The problem is that it doesn't scale. We already have "required",
> > "readonly", "writeonly" and people will likely want to extend it with
> > cardinality etc.
> 
> No no, it scales as good as "required", "readonly", "writeonly".
> You'd need an equal number of properties; they would just have a
> different domain;
> i.e., hydra:Class, not hydra:PropertyRestriction.

But you would need to attach them to every class/request template even if
the same definition is shared by all of them. Instead of reusing a single
definition.


> > So sooner or later we will need a "proxy" anyway. In this
> > case, I find it better to anticipate it from the beginning as
> > extensions requiring it are very likely to happen.
> 
> I'm still in favor of sticking a name to such a proxy;
> but if we have difficulties to find a name that doesn't contain
> "property",
> I'm afraid it will always remain tricky.

Me too... but I probably don't find SupportedProperty or, alternatively,
PropertyDescription as confusing as you do. This is not saying that is
perfect, you have a completely valid point.



--
Markus Lanthaler
@markuslanthaler

Received on Monday, 10 March 2014 20:34:43 UTC