- From: Ruben Verborgh <ruben.verborgh@ugent.be>
- Date: Fri, 22 Aug 2014 23:43:46 +0200
- To: Hugh Glaser <hugh@glasers.org>
- Cc: Luca Matteis <lmatteis@gmail.com>, Linked Data community <public-lod@w3.org>
Hi Hugh, > Can you tell me id there is a pattern for the uri= style stuff, where you want everything the service wants to say about the URI, in any position? The current triple pattern fragments spec does not mandate this, but: - each response will give you the controls (links and/or form) to find the other patterns - the server is free to include more triples than asked for - future extensions (that are planned) can support this > And I guess that raises the question of bnodes as well. My answer to that is always: bnodes are Semantic Web, but not Linked Data. If a node doesn't have a universal identifier, it cannot be addressed. That might seem like the simple explanation—because it is— but it's the only satisfying answer I have found so far. > I suppose I am looking at LDF from the point of view of it is a way of specifying the invoking URI pattern, and what my services would look like if they were using such patterns to be invoked - although maybe that is misuse? You could do that; that's one way of looking at it. The important thing is that a client doesn't have to guess or know anything about the server. Just by getting one arbitrary response (fragment), it is able to retrieve any other. No URL hacking needed. Best, Ruben PS Something I didn't mention in the earlier mail: it does combine nicely with dereferencing. For instance, the URL http://data.mmlab.be/people/Ruben+Verborgh 303s to http://data.mmlab.be/mmlab?subject=http%3A%2F%2Fdata.mmlab.be%2Fpeople%2FRuben%2BVerborgh.
Received on Friday, 22 August 2014 21:44:21 UTC