Re: By way of example.

>
>Isn't the best path here for the server to provide the right
>cache-control directives to make sure that cached responses don't
>"leak" across users.  If done properly, then it seems like each user
>has their own logical view (or "access context") of the resources and
>collision doesn't happen.
>

This is what I've been trying to say.  The spec as written gives absolutely 
no guidelines which would prevent such occurrences.  I'm not really saying 
the URI collision is a bad thing, what I've been saying is that the group 
needs to recognize it for what it is and not underspec here by pointing to 
RFC 2616 conneg which assumes the negotiation is always for fixed.variable 
and never variable.fixed, which just leaves an aliasing problem not a 
collision condition.

-EJB