application/atom+json? and more on our converter

Sam Ruby's recent post (Application/Atom+JSON) has sparked some interesting comments.  Is there value in a standard way to represent Atom data in JSON data structures?  I think so.  I do think that we need to have a way for transparent pass-through of generic elements (which means some scheme for dealing with Atom extensions, which means dealing with at least some subset of XML namespaces).

In other news, we're  working on fixing the issues that Sam noted in AOL's Atom-to-JSON converter.  The good news is that most of these issues have to do with the underlying ROME library we're using, and so the fixes that Joseph van Valen is making will be contributed back to the community. 

It turns out that the same issues are present even if you specify Atom as the output format; while Atom-to-Atom isn't a very useful converter, it's awfully useful as a test case.  For example, the Feed Validator tells us that something is munging hreflang.  It's also dropping rel="self" and other links for some reason... all of these things look like simple bugs, easily fixed.  The converter actually does seem to pass through elements it doesn't understand for entries, but not for feeds, which looks like a simple oversight.

Tags: json, atom, ROME, conversion, validation