- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Wed, 13 Jun 2007 11:11:53 +0100
- To: GRDDL Working Group <public-grddl-wg@w3.org>
I believe we are down to two open comments, one of which is this one: http://lists.w3.org/Archives/Public/public-grddl-comments/2007AprJun/0069 I tried replying http://lists.w3.org/Archives/Public/public-grddl-comments/2007AprJun/0096 and http://lists.w3.org/Archives/Public/public-grddl-comments/2007AprJun/0097 A short summary of the issues is: a) How does GRDDL work in a framework passing XML messages? b) What about time varying resources === This is all couched in the commentator's general dislike of some of the design choices of GRDDL, and so it is hard to see the wood for the trees. In particular, the commentator's repeated insistence that we should have defined GRDDL in terms of representations rather than resources seems to me to be a red herring. === Just a little bit more: a) How does GRDDL work in a framework passing XML messages? There are specific technical questions, which the commentator poses as problems with the spec, but are more constructively read as: how do I do this in GRDDL. I believe I have answered the technical questions, but have not yet heard back, whether these address the comments. I perhaps should have had a round of: what I hear you ask is ...., and having agreed the questions, I could then have shown how the spec addresses this use case. b) What about time varying resources, isn't the rule to merge the GRDDL results of a resource going to cause problems with time (specific example was a weather site). I pointed out in http://lists.w3.org/Archives/Public/public-grddl-comments/2007AprJun/0097 that a GRDDL aware agent is not required to merge results over time, but the spec licenses such an operation. On reflection, I think this is a feature. I think if you are publishing a time-varying resource, with a GRDDL result, then the results should be mergeable to produce sense rather than nonsense. A particular GRDDL application can then legimately merge your results over time, to, for example, plot a graph of the weather pattern. This would mean that for that sort of resource, the GRDDL results should be appropriately modelled, probably with a timestamp somewhere in the representation. Jeremy -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Wednesday, 13 June 2007 10:12:21 UTC