Re: [getf] Node MUST process? (was: [GETF] okay, here's an updateddraftwith Henrik's option B)


I believe that we are in agreement. My point was that the
language proposed didn't say that clearly because the terms
used are (IMO) overloaded (does the reader *really* understand
the distinction between SOAP message and message?)

I was attempting to be more precise in the language so that
it was unambiguously clear what we mean to say.



Jean-Jacques Moreau wrote:
> Hmm... I think you are actually saying something quite different; at least this is
> not how I've understood Noah's original proposal [1]. Maybe it's the "need do so
> for others" which is confusing.
> IMO, the points he was trying to convey were (or at least my interpretation of
> them):
>   1. The SOAP processing model applies to a single message only, in isolation from
>      any other SOAP message.
>   2. There is no state, correlation or coordination at the processing model level,
>      even, for example if you are using a MEP which involves sending mutiple
>      messages in sequence, each subsequent message depending on the response to
>      the previous message.
>   3. Coordination/orchestration between multiple message is done at the MEP level
>      (in hypothetical new MEPs), not at the processing model level.
> I don't think this spec necessarily needs to say that a SOAP node "is not
> restricted from engaging in activities that do not involve SOAP messages".
> Comments?
> Jean-Jacques.
> [1]
> Christopher Ferris wrote:
>>Hmmm... seems to me that what we're trying to say is that
>>the entity which we call a "SOAP node" is not restricted
>>from engaging itself in activities that are related to
>>messaging, but that do not involve SOAP messages per se.
>>So, it isn't clear to me that we have yet captured the correct
>>words to say this in spec-ese. (at least, I for one
>>don't read that in the proposals to date). How 'bout:
>>The processing rules defined in this section relate exclusively
>>to the circumstance in which a SOAP node *receives*, by (any|some unspecified)
>>means, a SOAP message. [note: I think that more correctly, we should probably
>>use the phrase: a message that conveys a SOAP envelope infoset" to be
>>precise] This processing is further qualified when a SOAP node
>>considers itself a SOAP intermediary node, acting in both the receiving
>>and forwarding roles, in that order.
>>Beyond that qualification, these SOAP processing rules have nothing
>>to say about the processing undertaken in the context of more
>>than one message (whether or not those messages convey a SOAP
>>envelope infoset) by an entity that considers itself
>>to be a SOAP node. The processing associated with related messages
>>is expected to be defined by a special class of feature that we call
>>an MEP (see section XX).
>>Jean-Jacques Moreau wrote:
>>>I would like to suggest the following further tweaks (changes
>>>between >>brackets<<).
>>><noah tweak="2">
>>>This section defines the SOAP distributed processing model. The
>>>processing model defined in this section applies to a single SOAP
>>>message >>independently<< of any other SOAP message; it makes no
>>>claim as to whether a given entity acting as a SOAP node for the
>>>processing of one (or more) messages need do so for others.
>>>While the specification of individual SOAP features (see 3.1 SOAP
>>>Features)>>,<< such as MEPs>>,<< may call for groups of messages
>>>to be processed in >>combination<<, this is >>independent of
>>>(orthogonal to?)<< the processing model defined in this section.

Received on Wednesday, 5 June 2002 09:24:18 UTC