Re: MIME registration for application/gdiff

>
>
> However, I'm not totally sure whether e.g. in HTTP, this should
> be used as a media type. What happens if some resource is negotiated
> by content type (e.g. like for
> http://www.w3.org/International/iri-edit/draft-duerst-iri)?
> How do the client and the server make sure they apply the diff
> to the same representation of the resource?
>
> Regards,   Martin.

The design of the PATCH headers, and whether to use a MIME type in the 
Content-Type field or whether to apply an instance manipulation, has 
already been discussed in the HTTP Working Group mailing list somewhat, 
so I'm cc'ing ietf-http-wg and bcc'ing ietf-types (in replying to 
Martin's mail on ietf-types).

I wasn't aware of any servers that actually allowed content negotiation 
between static representations.  I did think about this problem, but as 
far as I know there's no standard way of telling the server to create a 
new representation of an existing resource.  What happens if you have a 
resource that is in one Content-Type and you PUT a body that's another 
Content-Type?  I would think the correct action would normally be for 
the server to replace the resource with one of a new Content-Type.  How 
do you create a new representation of an existing resource in a new 
content type and how does that request differ from overwriting an 
existing representation with the new content type?

How does <http://www.w3.org/International/iri-edit/draft-duerst-iri> 
work?  Is it a dynamic resource where the different output content 
types are generated from XML by the server?  Or is it really a static 
resource that has several static stored representations of different 
types?  If it's the former, then there's no problem with PATCH as 
designed, as far as I can see.  If it's a static resource, then I'd 
need to know how those different representations are created and 
modified, as I'm unaware of any standard way of doing so.

Lisa

Received on Monday, 27 September 2004 16:37:26 UTC