Dear TC members,
I am terribly sorry that I didn't respond to any of
the earlier emails and did not participate in the
conference calls. The last few weeks I was working
on the manuscript of my dissertation and combined
with commercial work I did not have any time left.
The last few days I took the time to thoroughly
study the working draft. Good to see all the work
already done!
My most important feedback is:
- I like the distinction between the SOA model and
the service implementation model. However, the word
"implementation" is a bit confusing as people can
also interpret this as the internal implementation
of the services (internal design of the service). Of
course this is not what is meant. Maybe we should
alter this term (e.g. service technical
specification). Or we could give a better
explanation on what we do and do not mean with the
term "implementation".
- In my opinion it is important to make a generic
service repository that can be used for specifying
services made using different standards, e.g. Web
services, REST services, services based on Java RMI
etc. So I think separation of the specification of
functional aspects of the service and the actual
technical contracts (e.g. WSDL/XSD) is key. A lot of
information can be specified on a functional level
without knowing the standard used (e.g. goal of the
service, preconditions, postconditions, QoS, service
provider information). It seems that the repository
is focused very much on the WS-stack at the moment.
Or am I wrong about this?
- I think we need to make a clear distinction
between services that base their input and output on
the canonical data model of the organization or
those that do not.
- I think we are lacking some specification aspects
in the framework. These include: information about
the service provider (name, role/function, contact
information), pre- and postconditions (or effects as
they are called by the OpenGroup), an indicator
whether or not a XSD belongs to the canonical data
model or is a native XSD of some application, and a
data dictionary for specifying semantics of the
terminology used in the input/output of services. A
possible standard for pre- and postconditions is
Rule-ML.
- It's a bit unclear to me what the role of OWL is
at the moment. In the current working draft it seems
that OWL is used for building a service taxonomy. Or
am I wrong. OWL can also be used for modeling the
canonical data model, but I do not think this is
intended in the current document. But I can be wrong
about this.
I had a look at the list of issues in JIRA. If it's
OK with everybody I can contribute to the following
issues:
- SRAMP-3 RDF/OWL vs XML Schema
- SRAMP-18 UML Generation
Also, I can assist in modeling some of the
functional aspects that need to be specified.
This Wednesday I'll also join the conference call.
--
Met vriendelijke groet/Kind regards,
ICRIS B.V.
Consulting and Research
Linda Terlouw
Enterprise Architect
Mob: +31 (0) 6 24 380 962
E-mail:
linda.terlouw@icris.nl