RE: Web Service Description and stateful services - Debating on a) Stateful Web Service Instances b) Stateful Interaction - OGSI - Does web service support Abstraction? How?

>> -----Original Message-----
>> From: marco [mailto:events@oxfordsociety.org] 
>> Sent: Friday, June 20, 2003 5:36 AM
>> To: www-ws-arch@w3.org
>> 
>> I was hoping to have some feedback from 
>> architecture group members on (...)
>>
>> In particular, in relation to OGSI spec: 
>> Is it appropriate for the web service community to support a) 
>> or is that a wrong OO mindset that we should abandon?
>
> From: "Champion, Mike" <Mike.Champion@SoftwareAG-USA.com>
> Interesting question, but I don't know that this is 
> related to the question of statefulness.

Very interesting feedback is coming out here, with ideas on a 
WS-State language, on a WS-Coordination inspired one and with possible
candidate document verbiage.

Interesting feedback has come out from the Attributes Task Force too.
From the 23 Jun 2003 minutes:
"... stateful web services is being discussed in ws-arch 
... stateful interactions vs state on the server itself 
... OGSI tends to lean towards stateful instances 
... BPEL is more aligned with stateful interactions 
... should we clarify the use cases to distinguish stateful interactions
vs stateful instances"

But let me put my a)Stateful Service Instance b)Stateful Interaction 
question in the context I see it.
Does web service support Abstraction? How?

Different author gives slightly different meaning to the word
"Abstraction".
Here I want to express a meaning compatible with the following:
"Abstraction captures the 'general/specific' or 'example of' or
'instance of' structural relationship among objects, among functions, or
among states in the problem domain. [A.M. Davis - 1993 - Software
requirements]

"Software engineering is the application of scientific principle to 
(1) the orderly transformation of a problem into a working software
solution and 
(2) the subsequent maintenance of that software until the end of its
useful life" [A.M. Davis - 1993 ]
If the process of turning an analysed problem into a working software
solution is to be easy and effective, the software technology used to
implement the solution must support the primitives of problem analysis.
Abstraction is a fundamental primitive of problem analysis.

Abstraction is not equal to object-orientated approach. 
When I talk about supporting "Stateful Web Service Instance" I am
pointing at a possible way of supporting Abstraction, 
not to a full-featured object orientated approach.

Modelling and requirement languages often have the concept of
abstraction at their core.

-In UML abstraction is pervasive. UML classifiers describe collection of
instances. Typical classifier/instance relationship are class/object and
actor/scenario. 

-The Formal specification language Z is the one in which CICS was
written. In the Z language, the fundamental concept of Promotion is
based on the idea of factoring a global operation into a local operation
and a mixed one. 
The system is thought to contain multiple, indexed instance of the same
component. Local operation are defined on a single component instance
and then promoted on a global, system level.


So let's say that I describe the desired behaviour of a system in Z or
in UML and I use Abstraction.
Then I want to implement the system using Web Services.
And here comes the context for the a) b) question: 
Does web service support Abstraction? How?

Marco



PS:
a) might be used to provide b)
A web service instance could be the context of a stateful communication.
A message sent to a web service stateful instance, would be interpreted
in relation to that instance specific state. The state of the instance
would be the context for interpreting the message.
I am Not pushing this solution.
What I mean is only that "If a) was fully supported, we would
automatically have a form of b)"

In my dissertation I give personal answers and a possible solution to
the a) and b) question and to the related one "Does web service support
Abstraction? How?". Of course this is not the place to discuss them. 



Marco Adragna
Kellogg College
Oxford University
marco.adragna@kellogg.ox.ac.uk
Chief of R&D for Hit-Com

Received on Tuesday, 24 June 2003 06:48:33 UTC