RE: ebXML Representtion of Metadata

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

1) The XSL currently currently deals with the entire library of interfaces,
and is a one-time transformation.  This library would be registered, and
reference the original XMI document.   Yes, XSL is limited, and I currently
need multiple passes through the XMI and then the library to break them into
individual DTDs (Schema later).  XSL is only being used as a test, and it
certainly can be done with better in Java.
2) "Consistent" meaning that the production rules are an algorithm applied
to a model, and should generate the same result each time.  It is consistent
since I have a known target into the model, specifically the <<Interface>>
classes and from there I traverse.  Each operation on the Interface class
has an argument of a specified "type", which can be found in the XMI, and
each operation return is a "type" which also can be found in the XMI.  Each
type may be a class diagram or utility class, and the type name allows me to
find the starting class and traverse from there via the respective
associations to other classes (which from there does get into some ordering
issues, but only at that point).  Scott Hinkelman stated correctly that
these are NEW production rules, but similiar to those found in the XMI
specification.

Scott

Sorry for the delayed response.  I just got back into town from vacation, in
which I left my computer at home!

-----Original Message-----
From: Duane Nickull [mailto:duane@xmlglobal.com]
Sent: Wednesday, May 24, 2000 4:14 PM
To: Nieman, Scott; ebXML-Architecture List
Cc: ebxml-core@lists.oasis-open.org; Miller, Robert (GXS); 'Iyengar,
Sridhar'; David RR Webber; Cory Casanave
Subject: RE: ebXML Representtion of Metadata



>Also, we hope to generate the
>interfaces for an ebXML Registry and Repository based on our UML model
using
>XMI and XSL.
Are you sure you want to go down this path with xsl?  The performance hit is
really not worth it IMHO.  Can you elaborate on the interfaces you want to
expose?


>The biggest problems we saw with the XMI spec is 1) the mandatory
><XMI>..</XMI> tags which I am currently ignoring in the test XSL, as well
as
>2) ordering which I believe is a general problem with the model instance
>philosophy anyhow.

Scott, what is your take on whether  "consistent" results can be obtained
using this approach?  If ten people in ten different parts of the world do
the same XMI 1.1 meta data interchange from UML to an XML syntax, will they
get one result or ten different results for the same process?


Duane Nickull