UML and DSLs Again

Why Whitehorse doesn't use UML

Christian Nagel's OneNotes — Sun, 18 Apr 2004 00:32:00 GMT

Why Whitehorse designers are not using UML

Paul Andrew — Sun, 18 Apr 2004 06:22:00 GMT

Finally we got a clue...

.Sibman — Mon, 19 Apr 2004 00:23:00 GMT

re: UML and DSLs Again

Richard Gilyead — Mon, 19 Apr 2004 10:28:00 GMT

Service Designer looks very interesting. How will it relate to business processes? Will service invocations be referencable from business process models? How will this relate to the BPEL stuff in Biztalk?

re: UML and DSLs Again / COBOL

Greg Collver — Mon, 19 Apr 2004 13:38:00 GMT

I like many of the ideas present by David Gelperin at LiveSpecs Software
http://www.livespecs.com/modules.php?op=modload&name=News&file=index&catid=14&topic=&allstories=1
in his article "Precise Use Cases", but the one that stuck out for me was that he was greatly influenced by Demarco's "Structured English". I am thinking that pseudo-COBOL code would make great annotations to Use Cases.

re: UML and DSLs Again

Keith Short — Mon, 19 Apr 2004 17:44:00 GMT

To Greg: I loved the "Precise Use Cases" paper. To my mind this is a good example of a precise DSL.

re: UML and DSLs Again

Keith Short — Mon, 19 Apr 2004 17:46:00 GMT

To Richard: Our plan right now is to host a service (i.e. a special box on the Service Designer) that represents a BizTalk 2004 orchestration. This will allow you to design a collaboration of services as an implementation of a business process, some of which orchestrate others, as defined by underlying schedules.

re: UML and DSLs Again

Richard Gilyead — Tue, 20 Apr 2004 09:47:00 GMT

That sounds good as a way to invoke an orchestration (a "process in a box") but what I really meant was how do we specifiy the dependencies and business rules which control the sequence and execution of a series of services which contribute to the completion of a business process i.e the service consumer view rather than the service provider view? Is this a feature of the Service Designer itself?

re: UML and DSLs Again

Dan Fox — Thu, 22 Apr 2004 13:14:00 GMT

Keith, I'm very interested in these ideas but have yet to see what an actual DSL "looks like". Can you point me somewhere to see how what the artifacts you described in earlier post on the class designer actually look like? Thanks

re: UML and DSLs Again

Keith Short — Tue, 27 Apr 2004 21:32:00 GMT

To Dan: We are stil working on the tool-building tools. I hope to show a screen shot in the coming weeks of the tool we use here to build the metamodel of a DSL and generate extension classes for our underlying framework.

re: UML and DSLs Again

Keith Short — Tue, 27 Apr 2004 21:36:00 GMT

To Richard: If I understand your question well enough, the answer is "no". The tool used to actually create the BPEL schedule is the BizTalk orchestration tool that shipped with BizTalk 2004. What appears on the Service Designer is the Web service facade to the schedule. BTW, that tool was alos built as a DSL using the WHitehorse designer framework.

re: UML and DSLs Again

Horia — Mon, 03 May 2004 07:27:00 GMT

Interesting perspective. I've also read Alex Bell's article a while back and found his taxonomies hilarious and can appreciate where Keith is coming from when he's noting that UML 2.0 is bloated and still imprecise in various places. His example of the Service Connectivity DSL is quite credible, and I appreciate its intent. However, the case for a custom Class Designer DSL seems much less clear cut to me. Sounds a bit like throwing the baby out with the bath water. It almost looks like

Design is messy (very much like Buddha noted that "Life is difficult" when he "awoke"). It's clear that neither top-down nor bottom-up approaches make best sense in all cases. In truth, design ends up being a meandering, some form of journey between abstraction and implementation.

Design models seek to capture and highlight the significant choices while hiding away the rest of the complexity - that's where UML may still play a good role. Implementation models seek to capture sufficiently precise details and semantics to allow for code generation - that's where DSL's operate.

Many (most) developers are 'implementationists', and are naturally drawn to implementation designs (read code). Some developers are 'abstractionists' (not quite an extinct species yet), and are naturally attracted by design models, focusing on wider patterns, in a somewhat top-down perspective.

The big question to be answered is how relevant UML is going to continue to be and how much the DSL's are going to overlap with the various UML digram types. As always, time will tell...

Not all is lost I suppose since the DSL's and their meta-models have precise semantics, making them amenable to precise future digital manipulation if value will be found in that. I'm thinking of potential DSL<>UML 2.x/3.x transformations, and I'm guessing they should be feasible.

re: UML and DSLs Again

Martin Rosén-Lidholm — Mon, 10 May 2004 09:02:00 GMT

Steve Cook (Software Architect Enterprise Frameworks & Tools Group Microsoft Corporation) on DSL and MDA.

re: UML and DSLs Again

Martin Rosén-Lidholm — Mon, 10 May 2004 09:02:00 GMT

http://www.bptrends.com/publicationfiles/01%2D04%20COL%20Dom%20Spec%20Modeling%20Frankel%2DCook%2Epdf

Steve Cook (Software Architect Enterprise Frameworks & Tools Group Microsoft Corporation) on DSL and MDA.

Links about Domain Specific Languages

Alan Cameron Wills' WebLog — Tue, 25 May 2004 13:02:00 GMT

Links about Domain Specific Languages

Alan Cameron Wills' WebLog — Thu, 27 May 2004 14:37:00 GMT

re: UML and DSLs Again

DJ — Thu, 10 Jun 2004 16:56:00 GMT

Maybe the surveys show little use of UML by Microsoft Visio UML users because Visio UML is one of the least capable UML tools available. We tried using it for a while, but the fact that, unlike every other Microsoft tool available, you can't programmatically access the UML data, you can't do much with it.

Ian Mariano — Thu, 17 Jun 2004 16:20:00 GMT

re: On code generation from models

Stuart Kent's WebLog — Fri, 18 Jun 2004 14:52:00 GMT

re: UML and DSLs Again

James — Tue, 22 Jun 2004 23:01:00 GMT

Having read this thread through Grady, Simon Johnston, and the various Microsoft blogs, I'm surprised that the strongest reason for DSL's vs UML has not been mentioned in a comment.

The use of DSL's is a good way to increase the customer tie-in to the Microsoft tools and platform. (Not that there is anything wrong with that. :^))

Ian Mariano — Fri, 09 Jul 2004 18:23:00 GMT

re: UML and DSLs Again

dianying xia zai — Mon, 02 Aug 2004 06:54:00 GMT

[href=http://www.dmoz.net.cn/ wangzhidaquang]
[href=http://www.86dmoz.com/ jingpingwangzhi]
[href=http://movie.kamun.com/ mianfeidianying]
[href=http://www.kamun.com/ dianyingxiazai]
[href=http://music.kamun.com/ MP3 free download]
[href=http://www.pc530.net/ diannaoaihaozhe]
[href=http://www.5icc.com/ duangxingcaixingxiazha]
[href=http://www.dianyingxiazai.com/ dianyingxiazai]
[href=http://www.yinyuexiazai.com/ yinyuexiazai]

Grady Booch on verticalized tools and Domain Specific Languages

Vertigo — Tue, 03 Aug 2004 03:37:00 GMT

global::

Objects, Systems and Everywhere In-Between — Thu, 21 Oct 2004 23:22:00 GMT

Re: Whitehorse modeler != UML ?

Channel 9 — Sun, 01 Jun 2008 21:23:01 GMT

Keith Short has a nice blog entry on just this topic...

Keith Short s Blog UML and DSLs Again | Insomnia Cure

Keith Short s Blog UML and DSLs Again | Insomnia Cure — Tue, 09 Jun 2009 01:12:57 GMT

PingBack from http://insomniacuresite.info/story.php?id=9162

Keith Short s Blog UML and DSLs Again | Quick Diets

Keith Short s Blog UML and DSLs Again | Quick Diets — Wed, 10 Jun 2009 06:43:54 GMT

PingBack from http://quickdietsite.info/story.php?id=1585

Keith Short s Blog UML and DSLs Again | debt consolidator

Keith Short s Blog UML and DSLs Again | debt consolidator — Fri, 19 Jun 2009 17:28:48 GMT

PingBack from http://mydebtconsolidator.info/story.php?id=977

UML and DSLs Again

Why Whitehorse doesn't use UML

Why Whitehorse designers are not using UML

Finally we got a clue...

re: UML and DSLs Again

re: UML and DSLs Again / COBOL

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

re: UML and DSLs Again

Links about Domain Specific Languages

Links about Domain Specific Languages

re: UML and DSLs Again

Read More

re: On code generation from models

re: UML and DSLs Again

Read More

re: UML and DSLs Again

Grady Booch on verticalized tools and Domain Specific Languages

global::

Re: Whitehorse modeler != UML ?

Keith Short s Blog UML and DSLs Again | Insomnia Cure

Keith Short s Blog UML and DSLs Again | Quick Diets

Keith Short s Blog UML and DSLs Again | debt consolidator