Back and Forth on UML and DSLs

<cynic>Because they're making money on UML?</cynic>

Don Box wonders: What has baffled me for some time is why the UML camp puts so much emphasis on UML relative to MOF and XMI, which I&#8217;ve always found much more compelling. I prefer MOF as well, as I&#8217;ve written before, and I strongly believe Microsoft should have based its DSL approach on that instead of inventing its own...

*Sigh*. UML seems to be regarded with religious reverence in some quarters, which doesn't help objectivity.

People never seem to have a sense of humour when I enthuse "at least it's not as crappy as drawing Warnier-Orr diagrams".

"UML seems to be regarded with religious reverence in some quarters, which doesn't help objectivity."

And in other quarters, it seems to be held with religious contempt equal in its lack of objectivity. Perhaps people want desperately to hold onto a standard that has proven extremely valuable to the software community and aren't excited about another effort that doesn't even try to base itself on UML's underlying MOF, much less UML.

I think the lack of emphasis on MOF is because few have done what Microsoft is doing with DSL's, so there's been little to no support in the way of tools and such, resulting in little or no understanding of what MOF is all about. Microsoft could have gone a long way toward making that situation better by supporting MOF instead of going in a completely different direction.

Don Box is commenting on the ongoing debate between the MDA camp and the Software Factory/DSL camp. I&#8217;m not baffled &#8220;why the UML camp puts so much emphasis on UML relative to MOF and XMI&#8221;. It&#8217;s in the name ;-)....

I personally don’t see MOF as solution when defining modeling languages (even if they are UML-based). There are several reasons, but briefly: MOF covers the aspects of language specification only partially; it lacks some elementary expressiveness and is definitely too complex. MOF is also a standard which implementations seems to end up with something else than the standard MOF (calling it then XMOF, EMOF, MOF-like etc). If MOF is sufficient as a metametamodel, why that happens? Any suggestion?