[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: New Version Notification for draft-mehta-atom-inline-00



Nikunj R. Mehta wrote:


On Jun 10, 2009, at 11:52 AM, Martin Atkins wrote:



Nikunj R. Mehta wrote:
Based on feedback, I have split out the in-lining spec in to its own I-D. HTML: http://tools.ietf.org/html/draft-mehta-atom-inline-00 
Text: http://tools.ietf.org/id/draft-mehta-atom-inline-00.txt
I am also tracking open issues about this I-D publicly at http://code.google.com/p/atom-ext/issues/list. The source for this I-D is also available, if you are interested. 
Looking forward to comments on the I-D.
Is it actually necessary to support inlining of resources other than atom feeds and entries? It seems like this adds considerable complexity where I'm not sure that there are compelling use-cases.
I am glad you are speaking up about this. The original proposal for in-lining as expressed in draft-divilly-atom-hierarchy-01 was rather modest and supported a rather simple use case.
However, per Mark Nottingham et al [1], the use of the ae:inline is held up as the right way of extending Atom. Plus, per James Snell [2], the inline content's media type is better gleaned from ae:inline@type attribute. Also, per Al Brown et al. [3] [4] [5], it was more important to ensure maximum flexibility than demonstrate specific use cases.
To be clear, what I was suggesting was that if you were going to be pointing to the atom:content rules as the model for processing ae:inline then it would better for ae:inline to have its own type attribute distinct from the atom:link elements type attribute. How to handle things if there is a mismatch between the atom:link/@type and the ae:inline/@type (or the atom:link/@type and the actual content type of the linked resource after performing an HTTP GET) is a separate issue altogether.
Further, it is possible to define the "up" and "down" links generically, but to define -- separately -- some set of rules that limit their use to construct an Atom-only hierarchy. In a theoretical Atom-only hierarchy, "up" and "down" links that point to anything other than Atom documents would simply be ignored by a processor. Defining that Atom-only hierarchy, however, should not restrict the use of those link relations to build other kinds of hierarchies. 

- James