WebDAV properties vs. Atom extension elements

[Ezra Cooper <ezra@xxxxxxxxxxxx>]

Hi all,

I'm a little concerned about the direction which I see the protocol 
design team taking around extensibility of various objects. In short, 
the trend seems to be "Use WebDAV." While WebDAV is a great protocol 
with a lot of strengths and a certain amount of experience behind it, 
I'm uncomfortable going to others and saying, "Atom is easily 
extensible, just use this other standard."

XML is *also* a great standard with at least as much experience behind 
it as WebDAV. If we could accomplish all extensibility of Atom protocol 
objects (entries, categories, templates, users, etc.) just by adding 
namespaced XML elements, that could be a significant advantage. This 
way, rather than requiring two syntax formats that both client and 
server have to create and parse (namely, Atom XML and WebDAV-properties 
XML), we only have one. The burden on implementors is less, and Atom 
has a more convincing story about extensibility.

Implications

This would require an XML wrapper around templates. Admittedly, this 
increases the size of template transmissions, but only by a handful of 
bytes--and the upside is that we have an easy place to stick extension 
XML elements on a template.

A great weblog-authoring protocol could probably be created using 
WebDAV as its basic infrastructure: all metadata could be stored in 
WebDAV properties. But it concerns me that the WG is trying to blend a 
pure XML approach with a WebDAV approach, effectively doubling the 
burden on implementors (client and server) without much gain.

To put it concisely, I want it to be the case that WebDAV is not 
required for Atom. It's already been agreed that WebDAV should not be 
required for core Atom capabilities. Can we improve that and say that 
WebDAV is not required for core Atom, and not for Atom extensions 
either?

Thoughts?

Ezra