RE: Thoughts about path and intermediaries

> > orthogonal way.  That is, it should be possible for a sender to send the
> > envelope without knowing a priori the path it will traverse.  That leads
> > me to believe that we should use the word targetable instead of
> > addressable in the defi nition of intermediaries.
>
> Linking that with [1], the path may be a mix of implicit and explicit
> path, you may want to explicit a path to avoid doing discovery to go
> through a serie of firewall, for example, but you may hit some other
> intermediaries. Some intermediaries may be targetable (thanks marwan for
> this wording), especially if they need specific data to operate, like
> authentication.
> 2/ would be no, unless more data is needed for processing, there is no
>    need to have the explicit path (unless you want to count them, for
>    hops-limited messages -> see ping)
> 3/ yes, as the implicit path would be explicited there to require more
>    data for processing
> 4/ no, if the XML Protocol layer is not aware of this, the application
>    layer should have the same behaviour there.
> 5/ If needed yes (for user interaction during auth for example), but if
>    is hidden by the XML Protocol layer, it is also fine as long as the
>    relevant information is returned to the upper layer.
>
This would mean that neither the XML Protocol layer nor the application
layer is aware of an explicit path which would further imply that there
isn't an explicit path at all (or if there is nobody looks at it ;-). This
takes us back to a single hop protocol with all multi-hop functionality
pushed up to the application layer. Is this what you intended ?

Marc.

---
Marc Hadley <marc.hadley@uk.sun.com>

[1] http://lists.w3.org/Archives/Public/xml-dist-app/2001Feb/0082.html

Received on Wednesday, 14 February 2001 10:54:32 UTC