Re: [regrep] 7/29/2003: Conformance Efforts and ebXML IIC

["Michael Kass" <michael.kass@nist.gov>]

Farrukh and all,

 

   I've >snipped< portions of our previous conversation regarding ebRS conformance testing,

for brevity. Below are my additional comments highlighted in BOLD

 

----- Original Message -----

[MIKE] - I see an ebXML RS conformance test suite specification document,
>which contains a list of conformance test requirements, and a conformance test suite

>Also included  is supporting test material like that put together
>by Len Gallagher in the initial ebRS conformance test suite

[Farrukh]  Your detailed responses have clarified things very much and I feel we
are on the same page.

[Mike] I would suggest a "staged" approach, in which test assertions are
>compiled for a selected "profile" or "level"

[Farrukh]  +1 on this very pragmatic advice.

[Farrukh] Who produces automated test suite for verifying Conformance?

[MIKE] - The IIC is normally tasked with this effort.  Although again, due to our
>own resource limitations,we would ask for help from the registry TC in contributing tests.

[Farrukh]  I think that would be very reasonable. Is any one in the RegRep TC
available to start the effort by taking ownership of marking up the RS
spec for potential assertions to harvest into an initial skeletal CTS ?

>[MIKE] -  If you take a "divide and conquer" approach to RS testing, you may be
>able to make this easier.  For instance, choosing a single binding (e.g. SOAP) to write test requirements,
>
[Farrukh]  +1 on choosing SOAP binding as a starting point

[MIKE] selecting LifeCycleManager
>functionality as a higher priority for test requirements than QueryManager
>functionality...  could get the ball rollling sooner.
>
[Farrukh]  I think we cannot do only LCM and not QM. Maybe we should consider doing
More LCM and some QM.

 

[MIKE] - I agree.  We must test both.. but initially focusing on LCM might be best.

[MIKE] - I started "marking up" the ebRS 2.5 specification, pages 1-40, ( sectoins
>#1 - #6) I found that portion of the specification to be straightforward regarding
>identifying conformance test requirements.
>
>The LifeCycleManager section (section #7) appears to be straight-forward and
>easy to identify testing requirements, grouped into fundamental
>sub-categores of "parameter", "response",  "exception" and  "audit trail"
>types of testing
>

[Farrukh]  +1


[MIKE] The QueryManager (section #8) could be a very rigorous exercise ... I would
>suggest a "generalized" approach to FilterQuery or SQLQuery
>testing, in which "duplicate" query tree node testing is eliminated to
>simplify both test requirement writing, as well as actual test generation.
>

[Farrukh]  I think that tests will require publish using LCM and verifying using QM
that data written can be read back as expected.

 

[MIKE] - Agreed


[MIKE] Section #9, Content Management Services, is an optional service, that can be
>tested by creating/testing an IIC defined content
>management service to the registry, then verifying its conformant
>functionality.  This could be done by creation and publishing of a
>"dummy" validation service, I believe.  Because this is an optional feature,
>test requirements for this service could be described.. but
>actual implementation of test cases "could" be deferred if time/resources
>are an issue.
>

[Farrukh]  We could only do the default XML Cataloging Service initially which is
the onloy required part of CMS chapter.

 

[MIKE] - This is a good minimalist approach initially.

[MIKE]  Section #10, Event Notification Service, is also an optional service, that
>is well documented and fairly simple to generate test requirements.
>Acutal test case scenarios should be easy to construct.  Again, because this
>is an optional feature, it is a candidate for test requirement
>generation, but actual test cases COULD be deferred for a later time.
>

[Farrukh] I could see defering this initially altogether.

 

[MIKE] - Agreed


[MIKE] Section #11, Cooperating Registries Support is a complex scenario requiring
>federated registries for conformance testing.  The requirements
>are well stated, but implementation of test cases will be complex and
>time-intensive.  The specification does not eplicitly say whether this is an
>optional feature, although I assume that it is.
> 
>

[Farrukh]  Some features are required but most are optionall. I could see defering
this initially altogether.

[MIKE] - Agreed

[MIKE] Section #12,  Security an be a large exercise, particularly with all of the
>possible XMLDSIG options.  But this is a "required" feature, and can perhaps
>be described with a small initial set of test assertions and test cases, and
>a limited testing scope.
>

[Farrukh] Testing default ACcess Control POlicy is a minimal requirement IMO.

[MIKE] - Agreed


[MIKE] So from a deliverables standpoint, if you are doing the entire ebRS
>specification  I would suggest ( working in parallel if resources are
>available ) :
>

[Farrukh]  Indeed. That is why I ask team again to speak up idf you are bale to help.

[MIKE] 1) "Marking up" the ebRS document to identify testing requirements - 4 weeks
>

[Farrukh]  Maybe we can divide chapters among the peopel available.

 

[MIKE] - I will submit a basic guideline document to the RegRep TC outlining the

fundamental procedures and principles in compiling a Conformance Test Suite Specification

document.  We are working on that in the IIC.  This will assist your TC in getting traction

on initial requirements and test cases. 

[MIKE] 6) To create a "full blown" ebRS Conformance Test Suite 12 months
>7) To create a Conformance Test Suite Specification - (integrating test
>requirements, test cases and supporting documentation) - 1 month
>
>This is my estimate, based upon IIC work with ebXML Messaging Services.
>

[Farrukh]  Your estimates seem right on  to me and based on first hand experience.

Lets hope that some colleagues are able to volunteer time to begin this
work soon.

 

[MIKE] - I will try to re-arrange my schedule so that I can have Thursday afternoons free for

regrep conference calls, and we can discuss these issues and measure progress.  I will not be

able to make today's call, if there is one.

[Farrukh]  On a personal level I can commit to all review activity. I am reluctant
to take on any primary role I am over committed between spec work and
implementation.

[MIKE] - Understood.

Thanks again for laying this out so clearly for us to begin this
important work.

--
Farrukh