Re: XMI reality check

["J. S. Gangolly" <gangolly@csc.albany.edu>]

On Fri, 16 Mar 2001, Bob Haugen wrote:
...
> XMI seems to be the "standard", but I have also heard
> lots of complaints.
> This is a general call for feedback on XMI, from people
> who have tried it:

> * What UML tools have you tried XMI with?

I have tried Rational Rose and Together Control Center

> * Have you tried to take the same XMI model and
>   move it from one tool to another?

No. However, I have compared the XMI output from both
Rose and TCC on the same UML model, and they are sort of
comparable.

> * What problems did you encounter?

The biggest problem was the bulk of XMI generated.
I could have written better code manually, but then ...

Also, it would have been nice to generate schemas, but
the tools do not support that currently.

> * Were XMI models imported into a UML tool
>    really ugly, even if they might have worked
>    technically?

I did not import XMI import into UML. I was more
concerned with UML export into XMI -- the natural
application from our point of view.

The XMI stuff generated by UML tools IS grotesque.
But that is to be expected of any first generation
add-on.

> * Any other XMI comments...
> * What's a better format, if any?  (RDF?)

I was amazed to see the diversity of groups working (probably
in isolation). RDF, Topic maps (ISO/IEC 13250:2000),
DAML (DARPA Agent ML), XFI,...

In my opinion, this also relates to the debate on context that
is raging on the ebxml-core list.

It is also my opinion that context becomes crucial in data
exchanges at sufficiently high levels in aggregations, but
for XML data applications to coexist amicably, it is important to
isolate the contexts as objects in the UML model itself.
Perhaps we need to stretch (or enrich) the meaning of context to
accomplish this.

My interest is in the definition of context at high level data
exchange (such as financial reporting, XBRL), and I would
be very much interested in any experience people may have had
in that context (sorry for the language/meta-language confusion).

Jagdish
--
Jagdish S. Gangolly, (j.gangolly@albany.edu)
State University of New York at Albany, Albany, NY 12222.
Phone: (518) 442-4949  Fax: (707) 897-0601
URL: http://www.albany.edu/acc/gangolly