Questions from Brussels

["Petit, John" <jpetit@kpmg.com>]

What an excellent meeting that was! We were productive during the day and sated by beer at night (a rewarding combo!) However, at the end of the week in Brussels, there remained several outstanding questions in my mind. There are others, but the following stick out. I know that in the minds of many group leaders, there are answers or approached to these issues, but those solutions have yet to at be outlined. The issues I have here are somewhat different, or perhaps worded differently, from our “outstanding issues” section of the Architecture Doc. More than likely the answers to these questions have been stated, or I missed them in the specs. If anyone is inclined, I would appreciate a few answers to some of them.

Requirements

1. Does the scope of ebXML mean that we are defining a comprehensive and cohesive e-commerce system, or a series of independent, non-interoperable systems. Are we leaving the ability of interoperate at a global level off the requirements doc? I got opinions on both sides of this. The requirements document is rather vague on this (and most other issues).

Business Process Methodology

1. Are these UML process descriptions required? Can a company participate in ebXML without these UML process descriptions? Could XSL B2B stylesheets and rules be enough? I talked with some people from the BP group and there were varying answers.
2. Are we going to be defining a set of stereotype extensions to UML to accommodate process?

Technical Architecture

1. Many questions have come up in the course of our sub-groups work over the past six months. Anders made a good point in saying that these issues, which seem to be continually swept under the rug, should be handled more formally and thoroughly. What is happening with that?
2. The global search mechanism is still undefined. How will such a global search work? How will users find particular schemas to search upon? How will the perform common core searches?
3. How will items in repositories be updated when changes are made? What is the change mechanism?
4. Should we not have more UML designs in our document? Should these not form the basis of our architecture? If so, we should have them now. Perhaps we should also follow the Rational Method, as several other groups are doing.
5. If we are building this, or specing this, should we not follow the Rational Process and develop the domain model and  a complete use case list for requirements? So far we have not produced UML models of the architecture. This seems remiss as UML is playing such a large role in ebXML.
6. What about Autosearch GUI generation? I know that there will be some sort of facility to associate metadata documents with DTDs and schemas. How will these extra metadata documents be referenced or associated with the schema? Will this facility be able to accommodate documents that allow search description. This is a facility for agent control/both manual and programmatic. [NOTE: need to see how this facility will work, then perhaps make a separate specification and recommendation]
   1. I do not feel that this is out of scope for ebXML. Human interface is critical to effectively accessing the terabytes on information that ebXML will be dealing with. In order to access this information easily, we must provide a path by which schema authors can specify the interface that makes sense to accompany their schema. Such specifications will allow the automatic generation of multiple languages as well.
   2. If someone can tell me that this is beyond the scope of ebXML, or not worth addressing, then fine. If it has to be a separate specification, fine, but we should have the facility to work with it.
7. What about commercial exchanges? This should at least be addressed by example. What about other specs that should be included? OFX etc.? covered and gone over in scope.

Core Components

1. Will Core be defining a list of common business tags? This seems a real opportunity missed if they do not. If it does not, ebXML will not be able to perform cross schema searches on basic items such as price and business category.
2. Not addressed yet is the mechanism whereby a global search system will identify items in schemas that are called by attribute namespaces. This needs to be addressed for inclusion in the Tech architecture.

Transport/Routing and Packaging

Registry and Repository

1. What exactly are items in the repository and items in the registry? Clearly binaries will not be indexed and searched upon directly through ebXML. However, binary files will accompany XML documents (gif, jpeg, moov, applet, etc.). This needs further definition.
2. Who will control the registry of registries? What sort of organization can run a registry?
3. As this is supposed to be egalitarian, there are limited costs involved in this correct SO to RA?
4. How will the update of data be registered? What is the mechanism for this?
5. What exactly is stored in the registry? Will schemas and categorizations be part of registries, or will this be left up to the whim of the RA?

Technical Coordination

Cheers, John Petit
KPMG
XMLfs Team
Office: 970 728 9468
Mobile: 312 961 8956