Microsoft and Domain Specific Languages, Reprise

The Pharonic Architect is Blogging

Harry Pierson's DevHawk Weblog — Tue, 21 Dec 2004 06:50:00 GMT

re: Microsoft and Domain Specific Languages, Reprise

Mickey Gousset — Tue, 21 Dec 2004 08:38:00 GMT

Hey Jack!

I look forward to reading more of your thoughts, both on Mr. Booch's post, and Software Factories in general. I have been doing some research on "Software Factories", or at least how Microsoft is using the term. I think my biggest hang up, at this time, in grasping the concept, is the use of the word "Factories".

I look forward to learning a lot more!

Thanks!

Mickey

Another one of the team online

GarethJ's WebLog — Tue, 21 Dec 2004 20:33:00 GMT

Jack Greenfield responds

Aali's blog — Wed, 22 Dec 2004 10:58:00 GMT

Putting The Cart Where It Belongs

Jack Greenfield's Blog — Tue, 04 Jan 2005 04:41:00 GMT

Putting The Cart Where It Belongs

Jack Greenfield's Blog — Tue, 04 Jan 2005 04:50:00 GMT

The UML / DSL debate

Stuart Kent's WebLog — Tue, 04 Jan 2005 19:59:00 GMT

re: MDA again

Steve Cook's WebLog — Wed, 05 Jan 2005 15:53:00 GMT

UML, OOAD and MDD

Hartmut's Box — Fri, 07 Jan 2005 20:16:00 GMT

In addition to my previous entry, I'd like to comment on Jack Greenfield's view of UML and it's shortcomings in the field of Model-Driven-Development (MDD). Jack describes UML as a useful means of object oriented analysis and design here and...

re: Microsoft and Domain Specific Languages, Reprise

JD — Mon, 17 Jan 2005 08:54:00 GMT

Jack, what you describe about UML being more appropriate for documentation than execution was once undeniably true. In fact, when Rational Software was driving the standardization of UML (via the OMG)this was significant sticking point.

At that point in time, a counter proposal was hammered out by IBM and more ObjecTime, which already had executable models. The resulting compromise changes to UML (including rigorous definitions in the metamodel and powerful usage of sterotyping) actually addressed this concern of yours in fundamental ways. In other words, you describe UML 1.0 accurately, but UML 1.1 is a greatly deeper beast than you admit.

Of course, since then all the relevant parties (including ObjecTime and Rational) have merged together and now all are part of IBM. So it's clear why many in Microsoft view this as political in nature. Of course anyone promoting 'one true methodology' is also being quite political.

That said, your arguments do not go very deep here. It does not seem that you are truly arguing that the UML metamodel is unable to express the DSLs that you envision. (And if you are, then you should not simply slur it as a 'documentation' language, which is an out of date characterization)

Instead, you (and your colleagues) seem to be arguing that UML is 'beside the point'. This is not very compelling, as "Software Factories" and the new terminology being promulgated here seems equally 'beside the point'. Domain-specific languages are going to be built on some foundation. You can argue that it doesn't have to be UML, fine. But you will be providing your own conceptual framework and sets of tools to build them on. This is mostly political and not a fundamentally different approach (or not as it had been articulated, at least).

So, the following are some relevant questions:
1. Are you familiar with the latest versions of UML?
2. Do you deny that some metamodel is needed to underly the common aspects of DSLs?
3. Is there a fundamental limitation to the UML metamodel that makes it ill-suited as the basis for DSLs? [And can you articulate a precise limitation?]
4. Do you admit that there are already complex, complete UML models that are compiled, interpreted, and otherwise used for software execution?
5. Assuming that you are right about the unsuitability of UML (which orignated in software engineering) for modeling other domains, what commonality do you forsee between the DSLs? What are the key properties of the metamodel that you support?
6. How is the DSL framework (fundamentaly, the metamodel) that you envision different from the UML? [can you articulate a clear advantage?]

I hope that Microsoft develops a better, less academic approach than UML, but haven't been convinced yet.

[My bias? I actually am on both sides of this issue. I once developed for Rational and now develop for Microsoft. I personally love a good political war; like DCOM vs CORBA - I think both were ugly but DCOM was more practical. UML should not be the de facto anything, but I prefer to see criticism on the merits, not the politics.]

The UML / DSL debate

Stuart Kent's WebLog — Thu, 10 Mar 2005 14:42:00 GMT

TechReady 1 - 2005: Day 1 Sessions Update

David L's Blog — Tue, 02 Aug 2005 23:30:41 GMT

Ok, so Clarry and I where flat out yesterday with sessions, unfortunately, after each session, my stock...

Intel Software Network Blogs » Microsoft Joins OMG and Outlines Modeling Approach

Thu, 11 Sep 2008 09:03:58 GMT

PingBack from http://softwareblogs.intel.com/2008/09/10/microsoft-joins-omg-and-outlines-modeling-approach/