RE: [ebxml-iic-interop] MS interop draft and some test cases / pr ofile

[Jacques Durand <JDurand@fsw.fujitsu.com>]

hello Mike (and all),

one of my last mails before I go farther away.. (sorry won't take my laptop to the beach :)

 

The idea behind the Configurator is that, as Mike said:

- some new CPA may need by dynamically "installed" (i.e. made available in some form to the MSH, so that 

it will be used when any matching CPAId in a message requires it). (some implementations may not

switch that easily from one to the other...)

 

- I should admit that Configurator does not seem really useful, when (a) a predefined list of CPAs can

be made available to MSH initially, (b) it is easy for the MSH to switch from one CPA to the other, without

any special reconfiguration, just based on the CPAId, (c) no change or resetting (aside CPA) is ever required by MSH, when

a new test case is executed. Indded, Configurator is not exclusively for CPA, but was intended to also

support other impl-specific re-configuration, if applicable. (but then that would have to be defined...)

 

- but also: even if no need to do any config job once a set of CPAs has been deployed on an MSH,

teh Configurator action allows at least to do a preliminary "check" that the right CPA is there for each test case,

so that a problem here would not cause an undue and misinterpreted failure of the test case.

(should be more explicit in the doc I realize)

 

I do not have strong arguments in favor of keeping these "Configuration" steps in many test cases,

except these above, in case you don't see much use for them. The last suggested use could be replaced by

adding initial test case(s) that would exclusively check that all CPAs have been well deployed, and are accessible

once for all.(so COnfigurator action may still be needed, but does not have to be used at beginning of each test case).

 

regards,

 

Jacques

 

 

-----Original Message-----
From: Michael Kass [mailto:michael.kass@nist.gov]
Sent: Wednesday, July 24, 2002 12:59 PM
To: Eric VanLydegraf; 'Steven Yung'
Cc: 'steve.yung@sun.com'; 'prakash.sinha@iona.com'; 'ebxml-iic-interop@lists.oasis-open.org'
Subject: RE: [ebxml-iic-interop] MS interop draft and some test cases / pr ofile

Eric,

    Jacques will be on travel shortly, but will probably respond to your message, however I will add my understanding of this issue to the response:
I believe that the "Configurator" Action will receive an ebXML message with a payload containing a CPA, or CPA-like XML document
describing the messaging agreement between the test driver application and the test candidate MSH.   The candidate MSH will read in the CPA
payload and configure itself to operate in the mode described by the payload.  The messages that follow this step ( from both the test driver as well
as the candidate MSH ) will utilize the new CPAId and its implied configuration in their messages.  The test driver and candidate MSH are "bound" to that agreement. The testing driver may, however  alter its messages to break from that agreement in order to elicit what it expects to be the proper (error or otherwise) response from the candidate party.
    So the "Configurator" action will actually do both of the actions that you describe: alter the CPA payload document to reflect the changes in
messaging configuration, as well as create a new CPAId to reflect that change in subsequent messages.

Hope this helps,
Mike

   

At 11:59 AM 7/24/2002 -0700, Eric VanLydegraf wrote:

Overall, I think this draft is good and look forward to seeing the rest of the profiles detailed.
 
I do have a question on the section
Lines 341-347 Configurator
What is meant by "change the collaboration agreement for the conversation related to a test case or a set of test cases"
is the idea to switch out CPAID on an "Initiator" action for a test case or set of test cases or is the idea to actually change the contents of a CPA document itself refered to by a CPAID ?

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

From: Steven Yung [mailto:steven.yung@sun.com]

Sent: Sunday, July 21, 2002 11:22 PM

To: Jacques Durand

Cc: 'steve.yung@sun.com'; 'prakash.sinha@iona.com'; 'ebxml-iic-interop@lists.oasis-open.org'

Subject: Re: [ebxml-iic-interop] MS interop draft and some test cases / profile

Hi All,

Sorry this took so long.  I have taken a first crack at the various profiles for Interop test, please review (for the purposes of the draft, I have indicated where it matches the Drummond test, these comments will be removed in the final draft, but a statement will be place into the document to acknowledge the contributions that the Drummond Group have made).

I will continue to work on this document this week and hope to have another draft out before the end of the week.

Cheers,

Steve

--

Steve Yung, steven.yung@sun.com

Market Development Manager

Web Services/XML Development

Sun Microsystems

Phone: +1-650-786-8469

Fax: +1-650-786-3674

Jacques Durand wrote:

Steve, Prakash:

Here is an updated draft of MS Interop test Suite doc.

Note that I have not merged yet all material from the earlier draft of Prakash.

(I leave it to you guys to do that).

things to look at in particular:

1- We need to define some Interoperability Profiles (I only defined one very basic, to review/extend).

(the "interop profiles" we discussed this morning on the call.)

I have assumed so far a very basic interoperability profile (see Section 4, "IP 1"),

that mostly checks the ability to exchange various types of messages.

(Now, should some security, e.g. DSig, etc. be added? maybe.)

Some possible approaches when defining profiles:

- only one transport is adressed in a profile (HTTP, SMTP, HTTPS),

(so it could be we replicate our HTTP profile suite for SMTP?)

- as focus is on interop, we assume conformance tests have been done before.

(but some conformance tests may have an interop aspect:

e.g. proper Ack generation is a conformance test supposed to be done before,

but ability to receive an Ack from another MSH and process it correctly is an interop issue.)

- base profile(s) should only require core features in MSH.

- its OK for a profile A test suite to include/refer to another profile B test suite (if A extend B)

(e.g. secure profile extend base profile?)

- we could have a "standard" profile, and an "enterprise" one that tests advanced reqs, like passing large payloads, etc.

2- Look at the test Case material (section 4)

So far I have described 4 Test Cases for profile IP1 (a small subset of Steve's test list)

Each Test Case is complete with test steps and message material.

Also See if the Test Service actions are clear enough and sufficient.

Then if its OK you can use these as examples to build other test cases.

3- Could you review that draft, possibly do some update ( refine / add profile definitions, some test cases,

possibly do more merging with Prakash earlier draft ) by end of this week?

Then I'll add whatever latest version to our OASIS site as a work in progress.

Regards,

Jacques

<<ebXML_MS_Interoperability_TestSuite.doc>>