RE: agenda issues for TechArch ConfCall

["Duane Nickull" <duane@xmlglobal.com>]

Hi Peter:

Thanks for the comments.  I will clarify some of these..

<peter>
I was under the impression that a goal of the ebxml work was to target SMEs,
but I
didn't see that explicitly referenced in the material duane sent out. Should
it be?</peter>

The work I sent out is working towards a goal of encompassing everyone, from
the smallest ma & pa shop to the larger corporations (even corporate America
:-) ).  I didn;t explicitly reference either trading partner as a member of
these categories and left it generic.

<peter>
Also, I'm unclear as to why there should be two APIs, one for machines and
one for
humans. </peter>

The human interface is for the SME's.  The smaller companies may not be able
to afford a high priced machine API nor have the technical expertise to
implement such functionality.  Also, there could be an instance where a
machine looks for a repository item and doesn;t find the exact match but a
human could then perform a secondary search and find a similar item.  (maybe
version 1.16 of a regrep item vs. version 1.165.)  This would allow the
human to make a decision to re use the component and modify it suitably.  It
is also a good back up to a machine API to be used by people looking for
regrep items to decide:

1. whether or not it exists
2. can they reuse another parties regrep item

If people can;t search for these items,  we may have many people duplicating
each others efforts.

<peter>
The system represented in the slides seems to me to require a large
technical
infrastructure on the part of those using it to do all of this template
mapping.
This seems to me to be inimical to the goal of targeting it for SMEs.
</peter>
The key word here is "require".  From an SME's standpoint,  they don't
actually require the infrastructure to build mappings.  These are stored in
repositories (plural).  An SME can search for and retrieve the semantic
information models they need via a machine or human interface.


<peter>
>From the material, it seems like there is only one reg-rep. This seems to
me to be
impractical because it would have to be one big honking reg-rep....</peter>

No.  This was not the intention.  There would be lots of registries and
repositories to hold many different types of objects (both cached and
dynamic).  I used a simplified approach to demonstrate a transactional
process flow.

The deployment of registries and repositories will be defined by Scott
Neimans ebXML group.  I would suspect that he has lots more to say on this
subject as he has already forwarded some great feedback from this PPT.

I agree with the rest of your analysis on Regreps.  They could be deployed
and maintained by industry vertical groups, specific organizations, groups
like oasis, W3C etc.  There will also likely be an ASP model whereby SME's
can "rent" a small amount of space on a shared Reg-rep for their own
purposes.

Peter- thank you for the great feedback.

Duane Nickull