RE: Role of intermediary

Hi Marwan,

There is some discussion of whether the header/body distinction is really
useful [1]. In the context of what I think Michah was raising, I'm not sure
that the distinction is really useful either.

You can thing of a chain of processing elements strung out between sender
and reciever. This chain is segemented into short chunks by the partitioning
of these elements amongs sender, intermediaties and receiver. Together you
can think of this chain performing some transformation a message passing
down the chain. Steering through the chain might be dynamic, it might be
prescribed in some way by the sender, it might not be the same between
successive messages sent between the same sender/receiver pair... we have
settled none of these questions (that I am aware of).

I think the issue that Michah was drawing our attention to was the
relationship between the order of elements in the whole chain and the
transformation applied between sender and receiver. We could reorder
intermediaries and move segements of the chain around and/or we could alter
the order of module processing within a single intermediary.

It would be nice for the transformation to be completely independent of
order modules are processed in. Everything (at least end-to-end) would be
nicely orthogonal... it's not clear to me that its possible to do in
general.

There is also the question of the partial transformations seen at each
intermediary... indeed at each processing element in the chain. These
partial transformations (from receiver to current point in the chain) will
in general depend on processing order.

Stuart.
[1] Thread beginning at:
http://lists.w3.org/Archives/Public/xml-dist-app/2001Jan/0110.html



> -----Original Message-----
> From: marwan sabbouh [mailto:ms@mitre.org]
> Sent: 25 January 2001 13:57
> To: Williams, Stuart
> Cc: 'Lerner, Michah, ALSVC'; 'xml-dist-app@w3.org'
> Subject: Re: Role of intermediary
> 
> 
> Michah
> "In that regard I would expect it to try to take a position 
> with repect
> to the semantic properties of chains of
>             intermediaries, where there are questions about 
> whether the
> relative positioning of XP blocks in messages and the order that
> messages
>             (and blocks) are processed at intemediaries semantically
> significant or not to the overall operation performed by the 
> chain (and
> to
>             the partial view of the operation seen by the 
> intermediaries
> the message passes through)"
> 
> 
> My understanding is that Intermediaries only process headers 
> and not the
> body of an XP message.  Right?  Is that what you meant by : 
> "...and the
> order that messages (and blocks) are processed at intemediaries
> semantically significant..."?
>

Received on Thursday, 25 January 2001 09:22:44 UTC