Martin Atkins a écrit :
Nikunj R. Mehta wrote:
Recent discussions on the atom-syntax mailing list has revealed
interest in creating a new WG to look at several extensions of Atom.
Here are several extensions that have been discussed:
1. Hierarchy operations including meta-publishing
2. In-line representation
3. Bi-directional text
Please respond to this email if you are interested in helping to
explore a new WG further. Please also identify if you have a specific
interest area to consider for a potential new Atom WG.
I am particularly interested in the in-line representation part of
this. I'm currently working with a small group on an Atom extension
for describing activity streams[1], and frequently we've received
feedback from implementors that they want the pertinent content to be
inline in the feed to avoid additional fetches.
-----------------
We were in fact discussing recently the use of inlining atom:link
content to refer to external resources without necessarily requiring
an additional fetch. I think this model is similar to atom:source in
that it provides some non-authoritative metadata that may be
sufficient for display purposes while providing a mechanism for
processors to retrieve the authoritative metadata when necessary.
The model we were considering was as follows:
<link rel="related" type="application/atom+xml" href="...">
<entry>
<id>tag:example.com,2009:123545</id>
<title>Some Related Entry</title>
...
</entry>
</link>
To my mind, the model here would be that if
type="application/atom+xml" then the link element MAY contain exactly
one atom:entry or atom:feed element which provides a subset of the
information from the target document, much as atom:source does.
Why not using atom:content for that? The inlining you're presenting
really looks like it defeats the point of the atom:link element in the
first place.
- Sylvain
