ARe: Abstract Model contribution for Intermediairies (repost)

Dear Yuhichi-san,

Yuhichi Nakamura wrote:

> For 5, assume that a sender sends a request via gateway.  Others should
> know the gateway address, but do not have to know the
> address of the sender application.  Even in that case, the sender address
> MUST be recorded in the path record?

No, this is not what I was suggesting. I think the sender should not count as
an XMLP Intermediary; and so in your example only the address of the gateway
would be recorded.

> For 6, this is very good restriction from implementation point of view.
> However, a response is replied directly from an ultimate
> receiver to the initial sender, according to "XML Abstract Processing
> Model" document.  Any idea?

I am not sure which document you are referring to. If you are referring to Mark
Jones' "Abstract Model for Module Processing", then I don't think your point is
specifically addressed.

If you are referring to Stuart Williams (et al) "Abstract Model", then I think
you are misinterpreting what Stuart suggested. Stuart's intent was to allow a
response to either come back directly, or through one or more intermediaries,
hence the reason for the diagrams to look intentionally vague.

> Is there consensus how to address intermediaries (and handlers in the
> abstract model document)?

No, I do not think we have reached consensus yet. Mark Jones and I have been
floating our proposals around to try to move the WG forward.

Thanks for your comments.

Jean-Jacques.

Received on Wednesday, 14 March 2001 05:56:11 UTC