W3C home > Mailing lists > Public > public-linked-json@w3.org > September 2011

Re: Thoughts on framing

From: Ivan Herman <ivan@w3.org>
Date: Thu, 1 Sep 2011 10:40:28 +0200
Message-Id: <CAE20A9E-000D-4D40-BC23-A489CF0023F6@w3.org>
To: Linked JSON <public-linked-json@w3.org>
This is not an answer to one mail in particular, but my feeling now after having read through the thread of David, Danny, and also the pointers of Manu on framing. I think (I hope) I have a better feel now for what framing is, what it is good for etc. Thanks to all.

So... ok, I am 'sold' for the usefulness of this, as well as the necessity of having some sort of spec for this (ie, I retract my comment in my earlier mail). 

The main issue I have, and I think that is more important than the details of the algorithm, is the presentation. I saw that remark before somewhere, but we may end up with the HTML5 spec effect: we have a specification that goes into all the (necessary) details of implementations that, as Manu said in one of his mail, the publishers are, in fact, not interested in. That is all the more unbalanced because the first part of the spec has a clear tutorial style to it. Just as Danny and I were first pushed back by the algorithmic complexities (and, as RDF people, we are probably more ready to accept complexity than the JSON programmers/users) this will have a chilling effect on other readers. Ie, I think the document should really really be separated into the user/publisher document and the implementors' document. The definition, for the user, of the frame filters will be enough for complexity, we should not go beyond that.

It is another issue whether framing is something that would/should be part of the "main" features of JSON-LD, or should be part of an advanced section. The question is conformance: if I implement a JSON-LD processor, MUST I implement framing (and normalization) or not? And I have the impression that the answer should be no. If I implement a JSON-LD to RDF conversion engine, which is a perfectly legitimate processor to be used as an input to a triple store, then I am definitely not interested in framing. I would like to be able to call my processor conformant, though.

As for the framing algorithm, I was not masochistic enough (to quote Manu) to go through the details of the algorithm. But the issue reminded me of the concept of minimum spanning tree[1], Patrick Stickler's CBD[2], or others. What I do not know is how general we have to be in the framing filtering mechanism. But, clearly, you guys have thought through it more than I did...



[1] http://en.wikipedia.org/wiki/Minimum_spanning_tree  
[2] http://www.w3.org/Submission/CBD/

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf
Received on Thursday, 1 September 2011 08:40:52 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:18:32 UTC