Re: Orphaned annotations

On Mon, 18 Mar 2002, Jim Ley wrote:

>	 (Sean Palmer has also raised problems with any context
> information surviving HTTP content-negotiation
> http://someurl/#xpointer(...)  only applies to one representation of the
> resource, rather than the resource itself.

Yep (and a lively discussion that was)!  Fundamentally we should add
not only a time dimension (my original point) but also an HTTP
dimension before our pointers are reliable.

Dealing with HTTP is simpler, insofar as we can make the assumption
that two equivalent requests will generate equivalent responses.
We should however extend the protocols to keep a record of HTTP
negotiation headers[1], so we can be sure we're not annotating
totally different resources.

> Within the Evaluation and Repair group we've discussed checking when
> resources have changed and the extent of those changes - similar
> mechanisms I think are needed in Annotea so you know if a resource is
> likely orphaned through the document structure changes.

Yep.  I think I'll crosspost this back to ER, in case anyone from
our original discussion has more to say.

[1] simple test: where there's a Vary header, keep a record of the values
    used for negotiable headers in the request as part of a pointer.

-- 
Nick Kew

Site Valet - the mark of Quality on the Web.
<URL:http://valet.webthing.com/>

Received on Monday, 18 March 2002 15:35:17 UTC