March 30th, 2004

  • evan

lj namespace

I've been working on creating a version 2.0 of our RSS namespace, and I'm running into some snags.

It seems there are two applications for including LJ metadata in feeds:
  1. Providing aggregators/readers with more information so they can better display the posts.
  2. Providing Atom clients with the full data so they can edit and resubmit the post.
These two goals are different in slight ways. For example, the "preformatted" property is not of interest to an aggregator, because the HTML we provide in the feed has already been expanded by LJ (lj user tags expanded, newlines changed to br). More subtly, data like whether comments are emailed should never revealed to anyone but the author of the post, and are probably not of interest to an aggregator.

So my question is: does this call for two namespaces? Or should I make (1) a subset of (2), and just document that some attributes are not revealed? (In that case, should we include metadata like the preformatted attribute at all in (1)-style feeds?)

Additionally, to really support (2), LJ needs to provide the original post text in the feed with a mime-type of something other than text/html. Glancing over the atom API code, it looks like we don't do that anyway...