W3C home > Mailing lists > Public > public-hydra@w3.org > January 2015

Questions about new collection design

From: Dietrich Schulten <ds@escalon.de>
Date: Fri, 23 Jan 2015 10:16:47 +0100
Message-ID: <54C2117F.9090105@escalon.de>
To: "public-hydra@w3.org" <public-hydra@w3.org>
Hi,

some questions about the new collection design.

{
   "@id": "/alice",
   "collection": {
     "@id": "/alice/friends",
     "@type": "Collection",
     "manages": {
       "property": "schema:knows",
       "subject": "/alice"
     }
   }
}



1. When I look at an individual it is more difficult to see which 
properties it has, I find it difficult to spot that :alice :knows .. 
persons. But what worries me more: we do not seem to say that the 
property schema:knows has any value assigned to it at all. Will a 
reasoner somehow infer from additional statements in hydra-core that the 
Persons in /alice/friends are the value of :knows? If we really do not 
assign a value to schema:knows, I'm afraid we have not fully solved the 
problem.

2. As I read the example, the actual list of friends is a separate 
resource, so what we express here is a link to a json collection - as 
long as there is no :member property on the collection.  We could also 
embed the members (Alice's friends). But this poses the more general 
question: how does the client decide whether it should dereference to 
get the actual friends (or for object properties, the actual object). 
Checking for the :member property or any sub-properties that are 
expected for an object to see if another GET is needed seems weird.

Best regards,
Dietrich
Received on Friday, 23 January 2015 09:17:41 UTC

This archive was generated by hypermail 2.3.1 : Friday, 23 January 2015 09:17:42 UTC