News

UML 2.0 gets closer

• By Jack Vaughan
• July 11, 2003

ADT's Jack Vaughan spoke recently with UML guru Cris Kobryn as UML 2.0 was being prepped for OMG balloting.

Telelogic supported some UML 2.0 elements ahead of formal ratification. As UML 2.0 moves toward completion, do you ahve to change anything?

Not very many. In fact, a good analogy that works pretty well is I compare it to what’s going on in the wireless world for Wi-Fi 802.11 standard. If you notice, a couple of the market leaders there have been ahead of the pack in terms of implementing 802.11g. And while it’s not fully ratified, they’re confident enough that 802.11g for wireless will be ratified in the near future, and they figured that the delta between their implementation and final ratification could be done via flash-ROM upgrade.

Similarly, Telelogic has had a very aggressive view of the importance of UML 2.0 for our customers. Our Tau G2 products, which were released in Q4/2002, are extremely consistent with the specification that’s being firmed up now. So the amount of change that we expect will be required for our tools will be minimal.

In fact, I think that’s one of the most exciting things about working on UML 2.0 as opposed to UML 1.0. I think certain vendors, such as Telelogic, are going to be much more in earnest regarding implementing the standard to deliver new features for our customers.

Now, the things Steve Mellor talked about in terms of executable UML — did that get into UML 2.0?

Yes. One of the important new features of UML 2 is that we have integrated an action semantics into UML 2, so that full procedural semantics required for executable models are an integral part of the new standard. So for tool vendors, such as Telelogic, our tools produce production-quality code. That’s essential for doing that efficiently, consistent with UML 2.0, and that’s already in the language.

We feel that we’ve covered MDA from the point of view of the theory and also somewhat as a standard, and now we’re trying to talk with the tools makers and the tool users. What do you see?

Comments about UML 2.0 in the context of MDA: I look at MDA as an important conceptual architecture. A vision of how future, let’s say visual model-driven development might be; but it needs to be grounded. More than anything else, it’s essential that MDA requires the precision and the completeness that UML 2.0 offers.

I think to make the MDA conceptual architecture a reality, it’s really important that high-quality implementations of UML 2 get rolled out to users as quickly as possible. Otherwise, there’s the risk of, let’s say, some healthy skepticism that MDA is a conceptual architecture that doesn’t deliver on its promises. I feel that strongly.

What is your view as we reach the conclusion of this important standards process?

Obviously, the UML 2.0 effort took a bit longer than originally planned, but in terms of how things are wrapping up, you should note that not only have we done things like incorporate the action semantics I commented on that’s important for executable models, but we took upon ourselves some additional requirements, because we thought they were important to users.

The other important thing is, even though it was a fairly long and contentious process, at the end, the U2 partners- the submission team that I led- ended up setting an OMG record for participation. Over 50 vendors and power users of UML 2.o [participated], including, in the last phase, Borland, which is now a member of our submission team, and I think that’s noteworthy.