The plan at the moment is to continue with what we’ve got, which is:

- everything accessible by xml-rpc

- or also through a more ‘rest’ like interface, using GET and POST, with input url-encoded (so you can test in a browser) and output encoded in XML or JSON (the user gets to choose).

The usage recommendation would be:

- for general users, use xml-rpc unless you are particularly keen on xml, in which case use the REST interface with XML output.

- for laszlo/flash callers, use the REST/XML version.

- for ajax code running inside peepagg, use the REST/JSON version.

The big question in my mind is: should we make the xml version’s output look exactly like output from the flickr api, or should we make it consistent with the xml-rpc and json versions?

Flickr’s output looks like this (angle-brackets changed to normal brackets):

(rsp stat=”ok”)
[content here]
(/rsp)

whereas at the moment we have:

(response)
(success)true(/success)
[content here]
(/response)

I’m going to go look at flickr’s xml-rpc output now to see how it’s been done there - as they offer an XML-RPC interface as well.