Re: XMI reality check

[Philip Goatly <philip.goatly@bolero.net>]

----- Original Message -----
From: "Philip Goatly" <philip.goatly@bolero.net>
To: "Bob Haugen" <linkage@interaccess.com>; "'Kanaskie, Kurt A (Kurt)'"
<kkanaskie@lucent.com>; "'Jim Clark'" <jdc-icot@lcc.net>; "Welsh, David"
<David.Welsh@nordstrom.com>
Cc: "Race Bannon" <race.bannon@ia.com>; "'ebXML-BP@llists.ebxml.org'"
<ebxml-bp@lists.ebxml.org>; "'ebXML-CCBP-Analysis (E-mail)'"
<ebxml-ccbp-analysis@lists.ebxml.org>
Sent: Monday, March 19, 2001 11:10 AM
Subject: Re: XMI reality check


> Bob,
> I have tried the IBM toolkit below, and it falls over with any decent size
> model - i.e it runs out of virtual memory after about 45 mins ( Yes I have
> 416 megabytes on my PC unused !!). This did not fill me with great
> confidence I am afraid, since I can process the same model - every class-
to
> produce o/p to a database from which I can construct many different
> artifacts. The whole process takes 5 minutes with no memory problems. I am
> not using the fastest language available (VBA) but of course IBM is using
> Java on the PC - not I think a very good combination, if one wants
> efficiency.
>
> Still if the only tool you have is a hammer, the whole world looks like a
> nail
>
> Cheers, Phil
>
> ----- Original Message -----
> From: "Bob Haugen" <linkage@interaccess.com>
> To: "'Kanaskie, Kurt A (Kurt)'" <kkanaskie@lucent.com>; "'Jim Clark'"
> <jdc-icot@lcc.net>; "Welsh, David" <David.Welsh@nordstrom.com>
> Cc: "Race Bannon" <race.bannon@ia.com>; "'ebXML-BP@llists.ebxml.org'"
> <ebxml-bp@lists.ebxml.org>; "'ebXML-CCBP-Analysis (E-mail)'"
> <ebxml-ccbp-analysis@lists.ebxml.org>
> Sent: Friday, March 16, 2001 7:08 PM
> Subject: RE: XMI reality check
>
>
> > Kurt,
> >
> > Thanks a lot.  I assume this was IBM's alphaworks XMI toolkit?
> > http://www.alphaworks.ibm.com/tech/xmitoolkit
> >
> > Did you try exchanging XMI models between different UML tools?
> >
> > >IBM's toolkit was also better at preserving the model graphics when
> imported.
> >
> > Did that mean, preserving model graphics thru XMI was not very good
> > even with the IBM toolkit?
> >
> > -Bob Haugen
> >
> >
> > -----Original Message-----
> > From: Kanaskie, Kurt A (Kurt) [SMTP:kkanaskie@lucent.com]
> > Sent: Friday, March 16, 2001 12:26 PM
> > To: 'Jim Clark'; Welsh, David
> > Cc: Race Bannon; 'Bob Haugen'; 'ebXML-BP@llists.ebxml.org';
> 'ebXML-CCBP-Analysis (E-mail)'
> > Subject: RE: XMI reality check
> >
> > All,
> >
> > I have had some experience with XMI (Unisys's plug in for Rational and
> IBM's
> > XMI toolkit). I agree XMI is ugly but it was intended for machine to
> machine
> > exchange of models, not human readability. I have found IBM's version to
> be
> > more complete that Unisys's for what I was trying to do. IBM's toolkit
was
> > also better at preserving the model graphics when imported.
> >
> > I successfully used XSLT on the XMI from a UML model, based on Booch's
> > meta-model for XML Schema, to generate a human readable XML DTD. This
> > approach is being finalized by OAG as their method of generating DTDs
from
> a
> > UML model of all of their BODs (170+). Thus I believe that XMI can be
used
> > to generate XML instances of business processes modeled in UML. However,
> > this approach will require some consistency changes to the UML version
of
> > the Spec Schema.
> >
> > Another benefit of XMI is to generate meta-model instance DTDs that can
be
> > used to check model instance UML models. The stock tools to generate
these
> > DTDs are also quite ugly and not human friendly, but serve the purpose.
> >
> > Regards,
> > ________________________________________________________________
> > Kurt Kanaskie
> > Lucent Technologies
> > kkanaskie@lucent.com
> > (610) 712-3096
> >
> >
> > ------------------------------------------------------------------
> > To unsubscribe from this elist send a message with the single word
> > "unsubscribe" in the body to: ebxml-bp-request@lists.ebxml.org
> >