RE: Meeting Notes 02/13/2001

["Nieman, Scott" <Scott.Nieman@NorstanConsulting.com>]

Comments regarding the meeting notes:

1) I am pleased to see so many people turned out.

>>Comment one was that on the Registry Information Model on line 260 has
another box to show where the content would be stored. <<>> Comment one is
that Len thinks we need the ManagedObjectContent for clarification.

2) I am firm on my belief that a CONTENT does NOT need to be modeled. 
 
a) From a purist UML viewpoint, a model is synonomous with metadata (M1
level). To model content is to provide metadata about the content.  Len's
point is invalid, or more or less "inaccurate".  Even the MS File Allocation
Table (FAT) file management system does NOT model the content; it is ONLY
provide metadata ABOUT the context so you can locate it, and manage it.  FAT
is a simple Registry.  

b) Modeling the CONTENT is only required if we are building a file handler /
parser/ writer. Therefore, a content model should only be the format
specification of the file itself, which is important to business
applications that need to process this content.   Which leads to a point
that a registry is an application...

c) I believe the compromise that was done (intrinsic vs extrinsic) was
acceptable.   They reflect similar concepts, the intrinsic artifacts need to
be processed (parsed) since they are important to the functionality of an
ebXML Registry, particularily classification schemes.  

Farrukh's implementation provided a way to add schemes, which is not
specified in RIM.  The classification scheme needed to be an XML document
according to a non-specified DTD.  Perhaps this is a whole in RIM??

I think the Len's point is not separation of modeled content from the
metadata, but perhaps whether the registry has a PARSER available for that
content.  Perhaps the classifier is "parsable" or "indexable" - which, btw,
I mentioned in Tokyo.

>> Next Martin B. went through the metadata model they had for the core
components.   One of the key subjects discussed was the ability to query the
content and not just the metadata.  

3) Which gets to another point I made before, we should only allow content
to be searched if we can parse and index it.  Since we can store ANYTHING as
ManagedObjectContent, do you want to search through a bunch of MPEG movies
for anything that has an XML entity of <Party>??  Nonsense.

Scott