SD West 2007 gave me the opportunity to interview tens of different people of their initiatives in software development. The overwhelming majority said that they hate UML as it is not addressing any of their domains and does not translate the models to real code. UML is a generic language and tries to cover everything under the sun, but does not do anything especially well. What is also important to understand is that UML does not generate real domain-specific code, only stubs and skeletons that somebody has to write and maintain.
What really made my day was a discussion with one of the original inventors of UML (worked with the three amigos) within Rational who stated that UML was never intended to take care of code generation... That was something that came around almost accidentally due to some entreprenours being innovative. We all know that it did not work out and now we are seeing the results of it... People are reluctant to use UML.... But what is even more alarming is that the ones that have decided to scrap UML, have not chosen a replacement to it.... That comment came from a well know authority that I met during the conference.
So, why is there a sudden interest in Domain-specific Modeling (DSM). Let me try to summarize what I learned from the audience and by speaking to lots of very smart people:
1) People have a need to decrease the complexity to address ONLY the needs of the problem domain, not the problems of the entire world as UML sometimes has a tendency of
2) People need a way to sync their domain models with the implementation (code) which increases reuse and quality of code
3) Every company has one or two very smart and effective developers and their skills in algoritims etc. has to be captured in one way or the other. Why not reuse the skills of the good ones to be used by the majority of developers.
4) There is a need of change among software development. I have not seen any big improvements in productivity the past 15 years.... Yes, we have component-based development, object-oriented development but until people like Mr. Martin etc. admit that object-orientation is not the only wholy grail to development, things will not change. Mr. Martin stated in July 2006 as part of a panel in Dr. Dobbs Architecture and Design World in Chicago that software development should still be a craft..... Go and tell this to Fortune 1000 companies and their executives.... They will not be happy....
I am excited that DSM is finally getting there and people "get it" when you talk about it.