Matthew,
I have a general idea, but I would need to go back to the WS-CAF docs to check details.
IIRC, WS-CAF does what WS-RF does but doesn't bind the context token to an endpoint address. This allows the same context _token_ to be recognized by multiple endpoints, and that supports transactions. However, it's still a valid use of WS-Context/WS-CAF to generate a context inside the only servcie that will recognise it. I.e., we can use a WS-Context token instead of a WS-Resource to denote an activity.
Again IIRC, WS-CAF has the context-management controls of WS-RF: i.e. the ability to negotiate the lifetime of the context. It hhas the ability to notify "end of context", but not the ability to notify arbitrary progress-metadata.
When I mailed yesterday, I was in a debate about how DAIS/OGSA-DAI fitted with WS-I, WS-I+WS-CAF and WS-I+WS-RF. The conclusion so far is that all of DAIS, including the stateful stuff, can be done with any of those options. The choice of plumbing only affects the level of support and ease of programming.
So far, I can see two open-source implementations of WS-RF and I've found none for WS-CAF (still looking,haven't searched very far yet).
If you like, I can look at deatils of async activities in WS-CAF. Any takers?
Cheers,
Guy
On Thu, 10 Jun 2004, Matthew Graham wrote:
>
> Hi,
>
> I have just read the article and I think it puts a nail into the WS-RF
> coffin. Do you have any inklings yet of how the asynchronous stuff would
> need to be amended to use WS-CAF?
>
> Cheers,
>
> Matthew
>
Guy Rixon gtr-at-ast.cam.ac.uk Institute of Astronomy Tel: +44-1223-337542 Madingley Road, Cambridge, UK, CB3 0HA Fax: +44-1223-337523Received on 2004-06-10Z23:59:16