RE: [USTF] high level classification

First, I am moving this to the public list since there does not seem to be
anything secret about the discussion.  If I'm missing something sensitive
here I apologize.

As I documented before, the present document has the following high level
structure:

<quote>
This document specifies a variety of web services usage scenarios. It is
organized roughly as:

S0**: Message exchange patterns, ie rpc, asynchrony, security, reliablity,
conversations

S2**: Event based message exchange patterns

S3**: System and other messages

S5**: Service Description above and beyond those in less than 5** numbers

S6**: Discovery.
</quote>

The current document also has anchors that follow the convention shown
above.

I personally don't really care very much what structure is used to classify
the usage cases -- it seems to me that yours is rather similar in spirit but
different in detail from the one in the current document.  What I DO care
about, however, are the URI's -- because I would like to refer to them in
the EDI-like usage ensemble (whatever we are calling it now) document that I
am working on.  Putting in a whole bunch of links to the more atomic usage
cases seems to me a useful thing to do, but it is also a lot of work.  And I
would prefer, if possible, to do it once.  That is, the idea that the URI's
might all be changing shortly is somewhat demotivating.

I would like to suggest that we just take the document more or less as it is
and run with it -- perhaps ADDING to the high level structure as needed but
keeping what is already there.  I'm not hearing any reason why the current
structure is not good, simply that there is another that is also good and
might even be a little better -- or might not -- we could probably argue
about that.  But given the amount of sunk investment in the current
structure, is such an argument really necessary?  Can we just live with what
we have?

Of course, if we do that we would need, at the least, to add a higher order
structure reflecting the two types of usage case we have been discussing,
whatever we call them.  You know, the assemblages like Amazon and EDI and
the atomic ones like "Fire-and-Forget".  And possibly some more Sn's, I
guess, to reflect additional categories of atomic cases.

-----Original Message-----
From: Hao He [mailto:Hao.He@thomson.com.au] 
Sent: Thursday, May 30, 2002 6:31 PM
To: 'w3c-ws-arch@w3.org'
Subject: [USTF] high level classification


hi,

Here is a repost of the email I sent last week.

As of today's discussion, we might want to change:

<business-point-of-view> to <business-usage-scenarios>

and

<technology-point-of-view> to <technical-usage-scenarios>

Hao

----------------------------------------------------------------------------
--

As discussed in last week's phone conversation, it was agreed to have some
high level classifications of possible usage scenarios. Such classification
aids better understanding and collecting of usage scenarios.

I've come with a quick list I can think of. I am sure that you will throw in
more and better ideas. 

I tried to look at the classifications from business point of view and
technology point of view. Hopefully, we can combine them and get better
coverage of all possible and distinct scenarios.

<classification of="WS usage scenarios" >

<business-point-of-view>

<role name="Service consumer">

1. Service discovery

2. Service negotiation (Service level agreement, service attribute)

3. Service engagement

4. Service feedback, (Optional)

</role>

 

<role name="Original Service Owner (OSO)">

1. Service management (ensures the well-being of the service)

2. Service advertising

</role>

<role name="Service Integrator" >

1. Service integration

</role>

</business-point-of-view>

<technology-point-of-view>

1. Stateless services

2. Stateful services

3. Service identification

4. Service communication

</technology-point-of-view>

</classification>

Received on Friday, 31 May 2002 10:48:02 UTC