MDA again

I told Stefan Tilkov that speaking as an ex-participant of the OMG circus I particularly enjoyed the following paragraph from his blog:


So in the end, it seems to come down to MOF in its sort-of-working semi-compatible non-standardized incarnation (EMF), embedded into Eclipse, vs. Microsoft’s proprietary DSL stuff, embedded into Visual Studio. And the strangest thing is: I’d rather place a bet on something that is backed by than on an OMG standard nobody gives a f*#& about. Care for another example? Take this.


In response he suggested that I might not be so pleased by an old link in which he accused me of Microsoft FUD about MDA.  I guess I hit a few nerves with that particular article.


I was on the MDA panel at OOPSLA this year, where I identified the three MDA camps, and we saw that almost nobody in the audience was doing MDA.  I guess it hasn’t taken off yet.


Comments (4)

  1. I’ve calmed down a lot since then 🙂

    I know quite a few a projects that do what I would call MDA (although one may or may not agree with me that what they’re doing is *really* MDA – maybe it’s more appropriate to call it "MDA light" or "Pragmatic MDA" or whatever). In fact, I have been involved in a few of them (and I know of many more where I’m not). They do use UML do drive a model-driven development process. They gain significant, measurable benefits from it. I find that more and more often, I don’t have to convince people that MDA is worthwhile. Why is the perception so different?

  2. Steve Cook says:

    Another thing I said on the OOPSLA panel was that there are some people making a success of each camp. So maybe the perception is not so different. I think the key issue is whether UML is either necessary or sufficient for our objectives, not whether it is possible to make it work. Jack Greenfield has something to say about this in his blog at

  3. I don’t think we’re going to agree on the UML issue (which is OK). Can you elaborate on why you reject MOF? Isn’t it as good as platform to build DSLs as any other, with the benefit of being a standard?

  4. Steve Cook says:

    MOF isn’t a platform. The value of MOF in practice is XMI interchange. For example, few are interested in the CORBA mapping for MOF.

    XMI interchange is clearly on our roadmap for the DSL tools, although I can’t say when. We support it today with Visio. Of course, there are many different versions of XMI, and we have to choose which to support.

    We also want to provide a more flexible mapping from models to XML, because hardwiring the XML to the metamodel inhibits metamodel evolution.