Model driven development

[phitsc]

Anyone of you guys had any experiences (good or bad) with model driven development (using tools like IBM's Rhapsody)? I'm interested in aspects like efficiency, mastering tool complexity (learning curve), round-tripping, developer acceptance, management acceptance, comparison with CAD, etc. whatever comes to mind.

I'm also interested in opinions of people with no MDD experience.

[mouser]

my quick uninformed general impression is that these kinds of things become useful when you are working in a large team.  for a single developer or a very small close team of just a few people, they are more trouble than they are worth.

[mnemonic]

I've experience of Rational Rose (the UML part of Rhapsody) working in a large team

Pros:

• Allows a standardised documentation method.
• Means that information should all be in one place.

Cons:

• Once it's part of a corporate environment, the methodology guys add a huge layer of complexity to the tool (which they don't have to use).  I'm very much of the opinion that diagrams are there to communicate processes, rather than having to be a stickler for methodology-based rules and complexity.
• Like every documenting system, it quickly drifts out of sync with the development and becomes useless.  In a perfect world, the logical models and use-cases should be kept updated, but no-one ever has the time or inclination to do this.

Like mouser says, they are overkill for a small team.  I don't think that you can beat scribbling random flow diagrams and bullet points on paper or a whiteboard and then archiving them with your mobile phone's camera.