Domain Specific Modelling. Is UML really the best tool for the job?

Reaction to Grady Booch on UML and DSL's.

Stuart Kent's WebLog — Wed, 26 May 2004 18:29:00 GMT

re: Domain Specific Modelling. Is UML really the best tool for the job?

Rod — Thu, 27 May 2004 01:54:00 GMT

Give it up
You are not Booch

Reaction to Grady Booch on UML and DSL's.

Stuart Kent's WebLog — Thu, 27 May 2004 12:50:00 GMT

re: Domain Specific Modelling. Is UML really the best tool for the job?

Steven Kelly, CTO, MetaCase — Thu, 27 May 2004 21:08:00 GMT

Stuart is indeed not Grady, but his points are valid. UML has a good set of semantics for describing O-O programs. It also has a notation, 90% of whose symbols are unrelated to the semantics they represent (the Actor stick figure being the most obvious exception). That, along with its ubiquity, is the reason people try to use it to model entirely different things - with predictably unsatisfactory results.

In domain-specific modelling, most of the symbols will represent something recognizable from the domain - giving benefits Booch too accepts. The main point though is in the rules - e.g. how are you allowed to connect these symbols - and their semantics: what does a given symbol or connection mean in the system.

Most views of DSM (some in Microsoft appear to differ) include the idea of generating full code automatically from the models, using domain-specific code generators. Industrial experience has consistently shown such use to be 5-10 times faster than current 'standard' practices.

The challenge is to make building DSMs, and their tool support, as quick and easy as it should be. Microsoft's envisaged Whitehorse SDK, Eclipse's EMF+GEF, and metaCASE tools all offer ways to do this, the last even without any coding.

Links about Domain-Specific Languages (DSL)

Fabrice's weblog — Sat, 12 Jun 2004 16:56:00 GMT

Links about Domain-Specific Languages (DSL)

Fabrice's weblog — Sat, 12 Jun 2004 16:58:00 GMT

On code generation from models

Stuart Kent's WebLog — Mon, 14 Jun 2004 17:36:00 GMT

New Architecture Bloggers

Harry Pierson's DevHawk Weblog — Wed, 16 Jun 2004 08:12:00 GMT

re: Domain Specific Modelling. Is UML really the best tool for the job?

David Webber — Fri, 18 Jun 2004 00:22:00 GMT

Stuart,

Exactly on the money. Check out VisualScript as an environment that allows you to build models and then create XML.

I've just used it to do BPSS models - and for the exact reason that UML cannot handle the richness and precision needed. And especially *context*. Context is the missing part to UML IMHO. These newer XML standards we're building over in OASIS have context mechanisms at the heart of them.

Anyway - check out the VisualScript model for BPSS - view the JPG, the XML, and then download and try the model. There's also a tutorial on using the model.

http://drrw.net/visualscripts/

Enjoy, DW

re: Good IASA Meeting

Harry Pierson's DevHawk Weblog — Fri, 22 Oct 2004 13:43:00 GMT

By Request: My Response To Booch's Doubts

Harry Pierson's DevHawk Weblog — Tue, 26 Oct 2004 02:40:00 GMT

The UML / DSL debate

Stuart Kent's WebLog — Tue, 04 Jan 2005 20:00:00 GMT

On code generation from models

Stuart Kent's WebLog — Thu, 10 Mar 2005 14:46:00 GMT