That said, let's read along... [NOTE: this post grew and grew and I
admit it doesn't really have much of a point to make about XMI --
it's more about How To Read The Newspaper. I can imagine Infoworld
publishing an occasional dissection as a public service: How A Press
Release Becomes A Press Article. Perhaps set to a snappy Schoolhouse
Rock tune... please take this in the late-night spirit it was
> XML ties together disparate development teams
> By Dana Gardner
> InfoWorld Electric
Sorry, Dana, I hope you're not responsible for this lamer headline.
It's utterly uninformative.
> Posted at 11:58 AM PT, Feb 5, 1999
> The Object Management Group (OMG) and a group of its members are
> building a specification which will provide application developers
> with a common way -- using the Extensible Markup Language (XML) --
> to build powerful applications using a variety of tools and object
> types, the OMG announced Friday.
> The OMG's XML Metadata Interchange (XMI) specification is nearing
> completion as a standard for storing and sharing object programming
> and design information, said Richard Soley, the OMG's chairman and
> CEO, in Framingham, Mass.
OK, so why is this being "announced"? There's been an inscrutable
PDF file up on the site for a long while now (see technical
discussion and URLs at
http://www.omg.org/archives/orbos/msg00702.html). What's the news
> This means development teams from different tool and object format
> backgrounds -- perhaps to soon include the disparate Component
> Object Model (COM) and CORBA/Java camps -- will be able to
> collaborate more easily in developing shared and widely distributed
> object-based applications.
> "XMI enables developers to choose best-of-breed software
> development tools and to communicate among different repository
> tools, thus speeding the software development cycles and enabling
> best practices --- a powerful combination," Soley said.
> XMI uses the Web to exchange models between tools, applications,
> and repositories to stimulate object and component reuse by
> enterprise systems architects, integrators, and developers.
> "XMI will give companies the assurance that they will be able to
> reuse meta information across pieces of large-scale development
> projects, including coding, data architecture, and data
> management," said JP Morgenthal, president of NC Focus, in Hewlett,
Here's the early at-bat from a "user", the obligatory "independent"
quote that any tech journalist has to have to separate him or herself
from the hacks rewriting press releases. Unfortunately, I don't think
much of NCs, NC Focus, or Mr. Morgenthal, after "educating" him when
last we met. To the degree NC hardware has anything to do with
component software development -- and to be completely clear, I'll
say nil -- I'll trust this opinion when it's less vacuous.
> Perhaps more importantly, XMI may also reduce proprietary meta data
> interchange formats and speed the integration of CORBA, XML, Java,
> and COM-based development environments.
I feel like a Mutual of Omaha wildlife special: here we spy the rare
red-breasted OMG Spin in the open.
Or at least a "hope."
> Microsoft, however, has characterized the OMG's XMI specification
> as an "attack on Microsoft," which has its own meta data
> specification, the Open Information Model (OIM) before the Meta
> Data Coalition, said Thomas Bergstraesser, product unit manager of
> Microsoft Repository and developer of OIM.
> "Having different standards for the exchange of data and meta data
> is foolish and would be a big disservice to Microsoft customers,"
> Bergstraesser said.
Now I see the 11:58AM insertion. Microsoft's characterization is
presented fairly high up in this piece, before the reader has any
evidence to judge this counterspin. Which leads to the
Kremlinological metaquestion, WHY is this indictment presented so
Further amusement: what the hell is the MetaData Coalition? This is
news to me, and I think I'm at least an order of magnitude more
interested in the area than the average infoworld reader. Is there at
least a standard tagline that the reporter should use to cue us in:
stds org? trade assoc? loyal to whom?
I'm glad Thomas gives a shit about Microsoft customers. There are
others, you know. It helps to be magnanimous, especially when
attacking an open standard.
Finally, why is "OIM" coming from MS Repository (aka very large
databases), rather than the mainline XML development teams at MS --
or has there been another palace coup? Rather than say, web platforms
group (amusingly known as Weblications, see the "Extreme XML" advice
column, http://www.microsoft.com/xml/articles/xml061598.asp ),
Frankston's XML-Data team (for better or worse), or Office/Access?
And what are they doing pushing a standard without an X in it anywhere?
> "The OMG would have liked to become the owner of OIM, which is part
> of the reason for the OMG attack on Microsoft," Bergstraesser said.
> Moves to standardize meta data interchanges with XML should be done
> in conjunction with the World Wide Web Consortium's XML Schema,
> Extensible Query Language, and Extensible Style Language efforts,
> he added.
How nice to crosspromote W3C! I wonder if Dana tried calling W3C's
crack PR staff to see what they think of being used a human shield
between two members (after all, OMG is one too).
As we continue to hear the counterspin, now OMG is attacking
Microsoft. I haven't seen any attack yet. Maybe that comes later in
Finally, how does one become the "owner of OIM"? This points up the
earlier reporting hole -- what the heck is MDC?
> "In the OMG world of XMI, [Microsoft] Office and [Microsoft]
> Exchange would talk [World Wide Web Consortium] XML [data], and
> Microsoft Repository and Visual Tools would speak OMG XML [meta
> data]," Bergstraesser said.
Well, Thomas, at least take the time to be quoted correctly. Calmly
repeat after any of the 400+ WaggEddies you might have chaperoning
you: it's OMG's XM*eye* and Microsoft's XML*dash*Data and W3C's
*nothing at all*. What the hell are you doing creating a W3C standard
that doesn't exist? it's a *submission*. That means XML-Data is still
None of those brackets should be there. Be kind to your journalists,
be quotable as is.
> In response to Microsoft's Bergstraesser, OMG's Soley said:
> "Microsoft is a member of the OMG ... [but] chose not to take part
> in the process leading up to XMI; I do not pretend to know the
> reasons. ... This effort is squarely in the context of efforts in
> W3C; it leverages them directly. ... How in the world could that be
> construed to be an attack on anyone?"
"The Rophynol is squarely in the context of Katie's drink: it
leverages her completely... how in the world could that be construed
as an attack on anyone?"
Clearly, both sides are willing to wrap themselves in the
black-and-blue W3C flag.
Nevertheless, "in the context of" is rather disingenuous. It *uses*
XML, which any damn fool can -- but I see no evidence XMI is on the
ciritical path of W3C work on, say, RDF (which IS the context of all
W3C metadata work -- except micropayment, which is another rebuke
(iff you're a member)).
At the same time, Richard's right. Nothing I've seen yet makes XMI an
attack on any accepted public standard. Sitting back and saying
people are attacking XML-Data when all it is is Microsoft's opinion
(plus keiretsu members ArborText, DataChannel, and Inso) means
they're attacking... closed systems.
> In essence, XMI brings together XML with the OMG's Unified Modeling
> Language and Meta Object Facility specifications. Thanks to XML's
> Web connection, XMI can either be stored in a file system or
> streamed across the Internet from a database or repository.
Let's break that paragraph up further: The first sentence is the real
definition of XMI and should have been in the lede. UML is currently
a diagramming facility, and god knows what MOF is, I haven't been
able to figure it out. But XMI is just a DTD for capturing those
files in XML format. To the degree it can be stored on the Web, big
> Analysts say little effort will be required to broaden the use of
> XMI to tap into both Microsoft-oriented COM objects and Java-based
> objects, including JavaBeans.
> "A minor effort could bring [XMI] into the NT/COM or Java worlds,
> and it's done. It should be possible without breaking a sweat,"
> said Tim Sloane, an analyst at the Aberdeen Group, in Boston.
> "People need this, and they will come."
Hmm. I just don't know how to evaluate this. But I didn't know
Aberdeen Group was the kind of technical risk-management and
development house that could give me a learned opinion on harmonising
such divergent data models. Nothing at
http://www.aberdeen.com/ab_company/about/about.htm says the staff has
PhDs in computer science... Instead they trumpet "They are quoted
extensively in information systems and business publications and are
frequent conference and seminar speakers". Tim IS the best match
they've got to this article, though: He believes "The two competing
models in the marketplace are Microsoft's DNA and
Internet Object Computing (IOC). "
I'm glad to hear that if it's quotable, he thinks such competition is
> XMI is being written by such vendors as Unisys, IBM, Oracle,
> Platinum, Fujitsu, Softeam, and Daimler-Benz, and supported by
> Rational Software, Sprint, Sybase, Xerox, MCI Systemhouse, Boeing,
> Ardent, ICONIX, Integrated Systems, Verilog, NCR, and NTT.
> IBM, Unisys, and Oracle are expected to implement XMI in their
> products as early as the second quarter of this year, OMG officials
> "With XMI ... corporate developers can develop system-wide models
> which can then be delivered to and used by the various
> implementation teams even if they are using different development
> environments," said Emilie McCabe, vice president for IBM's
> VisualAge tools marketing.
Ho-hum. How about a real user quote, so I have some insight as to
what developer complaint this will address? "Right now, I can't
export my UML diagrams from product X to my SCCS and hack it apart
with PERL to generate progress reports"
> The XMI plan, which passed the OMG's Technology Adoption process,
> is now up for formal technology adoption vote by the OMG
> membership, the OMG will announce Friday.
Aha! Finally, a news hook! Glad to see one in the last paragraph of
the article. Perhaps I can go to the horse's mouth.
http://www.omg.org/ (home page <TITLE> tag? "TITLE", of course. Sigh)
Going to the calendar of events, I see Dr. Soley just spoke at IONA
World in SF, which I'd bet is the REAL news hook, why this
conversation was on Dana's desk at all. Then, in the press room, the
real source of the fuss:
http://www.omg.org/news/pr99.html#xmi, including a Media Alert for an
XMI audioconference yesterday afternoon.
I see from the quote list that Tim Sloane and Morgenthal were offered
up as a pro-partisan by OMG PR -- as is McCabe, whose quote was used
as-is. I totally understand working on-deadline, but if you're going
to issue an "Update" then find at least one new opinion. Otherwise, I
might as well read C|Net...
In fact, let's turn to News.com... and what do I find but a *better*
researched story! I'm shocked, as a loyal infoworld user, I'm
> Software developers to get new XML standard
> By Wylie Wong
> Staff Writer, CNET News.com
> February 5, 1999, 5:15 p.m. PT
> URL: http://www.news.com/News/Item/0,4,32072,00.html
OK, so Wylie also got stuck with a lousy headline.
> The Object Management Group today said it expects to pass a new
> XML-related standard by next month that will allow software
> development teams to collaborate via the Internet.
The lede clearly states why this article went out today. Note that
it's posted *after* the OMG-sponsored teleconference and *after*
Dana's piece -- I want to spotlight that handicap right up front.
> XML, or Extensible Markup Language, essentially lets programmers
> exchange information over the Web. A new extension called XMI, or
> the XML Metadata Interchange Format, allows developers to use tools
> from different vendors to build object-based applications.
There's a better one-line description of XML, especially for a
technical newsaudience: "XML lets programmers create new markup tags,
making it easier to exchange structured information on the Web" --
say Why it makes it exchangeable.
The second line above makes it clear who should be aware of this
news. Simplistic, but effective.
> "The process is moving very smoothly. There are no votes against
> it," OMG chief executive Richard Solely said today. "We can expect
> implementations of this at the same time the final edited version
> [of the standard] is available."
Aha. Quote simultaneously lays out a subtext: OMG is *invested* in
making sure the vote goes smoothly and portraying a success to the
outside world. Fair enough.
> A group of 29 software companies, including IBM, Oracle, and
> Unisys, presented the proposed standard in November to the OMG, a
> technology standards body.
The same meeting referred to in the October note which was the first
URL in this message of mine: the very first hit you get if you search
for "XMI" at OMG.org --- implying the facts are on the table, have
been, and today's news *is* political
> The voting process will last through March. Out of about 100
> potential voters, 40 already have voted in favor and four have
> abstained, said IBM software architect Stephen Brodsky.
> Brodsky said XMI--which also could be used for data
> warehousing--will allow developers to build applications faster and
> less expensively.
Why should XMI be used for data warehousing? Perhaps this is a closer
clue as to why MS's chief warehouser is so pissed: what is a
programming-model-interchange file format doing masquerading as a
general-purpose data warehousing format? And why, then, independently
of the "official" XML Schema working group?
> During a conference call today, Unisys fellow Sridhar Iyengar said
> his company will add XMI support to existing products this quarter,
> while Brodsky said his company will announce products soon.
Fellow, with a capital F. Hey, Sri sends a chunk of change our way at UCI :-)
[other than that institutional donation link, let me be clear that I
personally have no stake in any of the players I am critiquing today]
> Giga Information Group analyst Teresa Wingfield said one problem
> with XMI is that the Meta Data Coalition, a coalition of 47
> software firms including Microsoft, is working on a separate
Aha, thanks! Wiley introduces MDC cogently! With a link! MDCinfo.com,
> In December, Microsoft joined the Meta Data Coalition and
> transferred the rights to maintain and evolve the Microsoft Open
> Information Model (OIM) to the group. The OIM defines a way to
> exchange data, using Microsoft technologies including COM as well
> as SQL and Java.
"exchange data" is a little vague; "exchange databases" is more
accurate for only four more letters. MDIS is a portable archiving
format for databases as far as I can tell from the TOC.
MDC turns out to be another purpose-built $2500/yr trade association.
It makes no mention on its pages of W3C that I can see. They do link
to a previous Infoworld article by Bob Trott; I'd have hoped the
editors of Infoworld Electric would have linked between these stories
as well - C|net did...
> Both the OMG and the Meta Data Coalition are careful not to
> characterize their separate efforts as competition. In fact, the
> OMG has invited the Meta Data Coalition to its next meeting in
> March, at which the software coalition will submit its data
> exchange technology to the standards body, said Katherine Hammer,
> cochair of the Meta Data Coalition. "The goal is collaboration."
So in the space of four hours, now both sides are being "careful"?
What a shock. THAT's news...
but for that kind of insider dirt, one might turn to another kind of
scandal sheet altogether like computergram
> Wingfield does not expect the meeting between the OMG and the Meta
> Data Coalition to be fruitful. "The probability of them working
> together is zero," she said, adding that the separate
> specifications probably can be bridged by the products, such as
> those made by repository vendors.
And here we have the clincher, the real value added, a more neutral
scoop than Thomas' original complaint to Dana. An independent
analyst, who was not on any OMG press materials, offered a forthright
opinion on the *business* case (rather than a technical judgment).
Just an average reader,
PS. Hey gang, Active Server Pages are migrating over from NT IIS to
Apache, in particular Solaris:
ChiliSoft releases beta of development software for Solaris
By InfoWorld Electric staff
Posted at 11:50 a.m., PT, Feb. 3
ChiliSoft on Tuesday posted on its Web site a beta version of
ChiliASP for Sun Solaris with Apache Web server support. The company
said it was the first time developers could create Active Server Page
(ASP) applications on Solaris using an Apache server. Microsoft has
adopted ASP as the application server framework for its Internet
Information Server on Windows NT. The beta software runs on Solaris
2.51 or 2.6, with the commercial release expected at the end of this
quarter. Support for Apache on IBM AIX will be available during the
first half of this year, according to the company. ChiliSoft Inc., in
Bellevue, Wash., can be reached at www.chilisoft.com.