Re: attn: chairs, editors - [was WebDAV properties vs. Atom extension elements]

[Robert Sayre <mint@xxxxxxxxxxxxxxx>]

Tim Bray wrote:

> 1. I've never really studied DAV.  I suspect that I'm in the majority.  
> It would be a disadvantage if someone had to grok DAV before being able 
> to implement the Atom protocol.

Certainly. We are already asking for more extensive knowledge of HTTP 
than most developers have. Asking them to understand WebDAV is out of 
the question. However, every developer understands path segments and 
separators, and the semantics added by DAV are straightforward to anyone 
familiar with file systems. [0]

> 3. If we do some or all of the APP in a DAV-compliant way, to what 
> extent does that mean that existing DAV implementations can be re-used?

DAV doesn't have a notion of creating resources on the server, though 
the lack of a "MKRESOURCE" action is often lamented. It also doesn't 
define GET for a collection. If we model collections on DAV, existing 
DAV implementations can be augmented with our requirements.

> 4. If we do some or all of the APP in a DAV-compliant way, to what 
> extent does that create expectations that Atom is a DAV application and 
> you should be able to hook up a DAV client and everything will just 
> work?  

Generic DAV implementations act sort of like FTP servers. This typically 
assumes Level-2 DAV compliance. One example of this is Apple's .Mac 
service (You can also connect to any DAV server through the Finder's 
"Connect to Server" dialog).

Another kind of DAV implementation builds on the model, but doesn't 
necessarily allow that kind of interaction. Lots of CMSes work this way, 
other examples include Exchange 2003 and Hotmail access through Outlook 
Express. The benefit here is the collection model of WebDAV: membership 
and consistency are clearly defined. For example, when the Etag on a 
PaceCollectionSyntax document changes, does that say anything about the 
collection, collection members or the other collection documents?

Robert Sayre

[0] http://lists.w3.org/Archives/Public/w3c-dist-auth/1998OctDec/0305.html