Re: Handling unknown elements in an entry

[Robert Sayre <mint@xxxxxxxxxxxxxxx>]

Dare Obasanjo wrote:

> > From the perspective as the person responsible for a
> lot of the server back end of MSN Spaces, it's a
> non-starter to expect that we'll store arbitrary XML
> sent from clients that doesn't map to properties in
> our internal data model. 

I wouldn't expect that. What about the set of elements defined by the 
Atom format? How about adding name/value pairs?[0] If clients press 
their luck, you can always return an error.

> I see your perspective but I don't agree with it. HTTP
> allows servers to ignore headers they don't support
> and still return 200, I don't see why Atom should try
> to be any different. 

My view is not set in stone. I'm just looking for a way to make this a 
little easier on clients. If the server's response is 100% dependent on 
its internal data model, that's not much of a protocol, is it?

Well have to settle on a small set of standard features[1] if clients 
are going to have a shot at implementing this, IMHO.

Robert Sayre


[0] 
http://atompub.org/2005/03/12/draft-ietf-atompub-format-06.html#rfc.section.6.4.1
[1]
http://inessential.com/?comments=1&postid=2984