RE: Conformance and naming

[Lynne Rosenthal <lynne.rosenthal@nist.gov>]

David

I don't disagree that the interfaces and interoperability are key. And as
you know, NIST is very active in developing the metrics (e.g., a Reg/Rep
reference implementation, XML and DOM Test Suites)  to accomplish
these key aspects.  HOWEVER, a specification needs to articulate
what it means for an implementation (e.g. system, software, message, etc)
to conform to that specification. Without this, it is difficult if not
impossible to ensure that developers of implementations will have a
common knowledge of what needs to be implemented to conform and thus, be
one step closer to achieving interoperability.  A Conformance Clause
in a specification states the requirements or conditions that need to be
satisfied to claim conformance as well as who or what must conform. 
Furthermore, without a conformance clause, it is difficult to develop the
test suites (i.e. metrics) to determine conformance - since the
conformance clause indicates who or what needs to conform and what needs
to be tested to show that conformance is met. Additionally, note that it
is common practice in ISO, IEEE and now W3C specifications/standards to
include a Conformance Clause.  

It is my understanding that many of the specifications as well as many of
the developers talk about conforming to ebXML or conforming messages,
etc.  It would be nice if we all had a similar or at least a
knowledgeable understanding of all that is involved when talking about
conformance.  We are not trying to dictate what needs to be said in
each specification, but rather are volunteering to lead a discussion of
the issues related to conformance - for example:  (1) will
extensions be allowed and if so, what impact does this have on
implementations that claim to conform; (2) will there be levels or
profiles of conformance and if so, are they
hierarchical and does a
higher level include conformance to a lower level.  It is up to each
ebXML Committee to determine what it means to conform to their
specification and what should be included in the Conformance Clause.

regards
Lynne



At 11:21 PM 10/16/00 -0400, David RR Webber wrote:
>Message text written by "Lynne S. Rosenthal"
>> 
>Hello All
>
>Conformance is an important topic and it is important that we all
have the
>same basic understanding of what conformance is, what the issues that
need
>to be addressed are, etc.  From previous discussions, it seems
that the TA
>spec should address the overall ebXML conformance as well as the
general
>principles that each of the specifications should follow.  
>
>NIST has been involved in developing conformance statements in
various
>specifications as well as developing other conformance related
products
>(e.g., test suites and test procedures).  Mark Skall and I have
volunteered
>to present a discussion on conformance at the Tokyo plenary.  If
we aren't
>on the plenary agenda, we would gladly present and lead a
conformance
>discussion during the week, inviting all who are interested (how
would that
>be arranged).  Additionally, both Mark and I are willing to meet
with any
>of the working groups to discuss the conformance clause and issues
for
>their specification. 
>
>Regards
>Lynne Rosenthal
>Conformance Testing Group
>NIST 
><<<<<<<<<<<<<<<
>
>Lynne,
>
>>From my view point - the key areas here are the interfaces - the
TRP and
>RegRep 
>ones, and the internal representation of the metamodel therein with a
>consistent reference system (UID's in the TA doc).
>
>Provided two systems can effectively interchange, and that the
information
>has
>a consistent information model - it will work.
>
>These are the areas I see that NIST led work would be valuable - but
we are
>
>not quite there yet in being able to exactly specify all this -
>probably post-Vancouver will be when this is stable.
>
>At least right now NIST can help collect the metrics required and
generally
>organize this so once we have the detail - we can quickly integrate
this
>into
>a sensible 'Implementors Guide'.
>
>Are these the right thoughts?
>
>Thanks, DW.