Re: XMI and UCI and you

Rohit Khare (
Mon, 08 Feb 1999 12:48:16 -0800

Yep, we've been discussing XMI and related work in the halls, and I did
notice your invovlement in the project -- but I sent it out just the same :-)

I walked through the MOF DTDs with Dan Connolly and I have a slightly better
understanding of it, but it took a step back to understand it: it's an XML
format for storing MOF data, rather than an "XMLized" data structure (meaning.
"it took a while to disentangle why all tags seemed to allow all other tags
in their content models")

Currently, for example, I'm reengineering the UCI C2 development
environment to migrate from an archoitecture derscription language as
EBNF to an extensible XML structure and facing the same dilemma: do I
lexically represent what was already in the old C2SADL parse tree, or
create fresh constructs and leverage hyperlinking to replace the old
style of lexical references? [And then, of coure , how can I let other
elements of the IDE annotate UML diagramming info and so on].

In that sense, I feel I'm crawling around the XMI problem space --
thus distressing me that I can't (yet) figure out if I need it, and if
so, how to use it (hack it).

I'm sorry we haven't met, but briefly, on your campus visits. Perhaps
we will have a chance to discuss these developments soon. For example,
it would be very exciting to bring the current industry perspective on
model exchange into Prof. Taylor's classroom...

Rohit Khare
626 806 7574



I have discussed XMI with Jim Whitehead et al at UCI. I did find your
e-mail commentary a fresh break from the other dreary e-mails. We are
discussing some joint projects with UCI already.

By the way the MOF (Meta Object Facility) is OMG's metadata repository
standard and is used to generate CORBA (Or whatever) interfaces for
programmatic access to metadata.


Sridhar Iyengar
Unisys Fellow