[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [soa-rm-ra] Service Interface Model (graphic attached)
Ken,
There's abstraction and then there's ABSTRACTION.
For example, at the RM level, it's OK to state that an
Information Model has Structure and Semantics; however, at the RA level, this is
not enough detail. This is why I show structure and semantics as
applied to messages and data (payload) within the Information Model
class.
In addition, the former visual model (Fig. 20) introduces new
concepts like Action Level Policy, Precondition, Service Level Interaction
Policy, Service Level RWE, and Action Presence. Concepts for which I do
not even know the source, RM or otherwise. Nor have we even had a chance
to debate the merits of such concepts.
Also in Fig. 20, Action is introduced and shown as associated
with Message with the label "invoked via". This is a somewhat dynamical
view and not really consistent with our RA notion that messages "denote"
actions (and events) and that actions "cause" RWE and events "report" RWE.
In the RM sense, we state that interaction is accomplished by performing actions
against the service and this is realized via message exchange (although other
"modes" are permissible, e.g., state changes to shared resources).
I think we all agree, based on feedback during the recent
telecons, that we do indeed need to revisit our discussion of Action
throughout the RA to insure consistency. I'm definitely in agreement
there. But for this discussion, I was making an attempt at proposing a
reasonable model for Service Interface, and one that would be readily palatable
by the reader.
- Jeff
|
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]