Re: [ebxml-cppa] Re: [ebxml-cppa-negot] RE: BSI distinguishingsuccess, failure,and transition conditions

[Matthew MacKenzie <matt@xmlglobal.com>]

Bob,

You're putting too much in IIC's lap with this suggestion.  The IIC 
process, of which I am an author, sees the IIC group extracting 
conformance requirements from the various specifications, then defining 
concrete test cases which reasonably cover all requirements.  We have 
enough work without having to help do specification work.  I don't 
think you meant your statement to be interpreted as I just did, but I 
assure you that the outcome of your suggestion would be something close 
to what I've stated.

As far as BSI being an "abstract" concept, and thus left unspecified -- 
I don't agree.  Who made the rule that "abstract" should not be 
specified?  In a sense, a concept is only abstract until it is 
implemented.  If the concept is not specified, then it will not be 
properly implemented.  Think of it this way:  The WSDL schema is 
abstract, but a WSDL instance is not.  I think that BSI should be 
specified and defined, because it will provide a serious reality check 
and touch point to all of this BP work.

Cheers,

Matthew MacKenzie
Chief Software Architect
XML Global Technologies, Inc.
p: +1 (604) 717-1100 x107
m: +1 (604) 781-4721

Attachment: Matthew MacKenzie.vcf
Description: text/directory

On Tuesday, Aug 20, 2002, at 11:51 America/Vancouver, bhaugen wrote:

> Re the periodic BSI controversy,
> here is a modest proposal:
> Start to develop a suite of conformance tests,
> in collaboration with the IIC group.
> Then people can start doing test-driven
> development of implementations that would
> satisfy the tests.
>
> -Bob Haugen
>
> P.S. I also think a design spec would be useful,
> even if non-normative.  This thread repeats often
> enough to be a sort of proof that it's necessary.
> On the other hand, I think that the eBTWG
> Business Collaboration Protocol project might
> come up with enough spec so that combining
> that with a test suite would be enough for most
> programmers.
>
>
> ----------------------------------------------------------------
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.oasis-open.org/ob/adm.pl>
>