I just sat through a presentation today that attacked Model driven Architecture. Now I am not enamored of MDA and generally (but not completely) agree with Scott Amblers analysis and training on some of the critical drawbacks on MDA . But I could hardly agree with 4 of the 7 major points used to attack MDA:
1)MDA is top down analysis gone berserk. This is the old political labeling, “this person is a liberal/conservative – horrid thinks can happen to you if you trust them”. Apparently, Analysis Paralysis is only one of the major blights on ITKind that MDA is trying to inflict on the world. But not to be accused of Pacman as Wild Bill Coding Cowboy, this vendor did allow for some utility for UML Use Cases.
2)Generation technology is in its infancy – and only we have the key insights to make it work properly. Yeah, right – and what are all those ASP, JSP, PHP sites doing ? Or go back one generation and look at all the Unix and mainframe code generation tools. And you want PC desktop generation tools – take a look at DOS Foxpro, Clipper, Paradox, Magic and a host of DOS desktop template driven tools.
3)Forward and reverse engineering is hard to do. No argument here; but UML and especially embedded processing toolmakers are getting awfully good at it. Factories for software development are not a unique concept and embedding the smarts for and reflections skills into a language are a part of both Java and C#.
4)Cross platform integration is over-emphasized. Wrong. Under-emphasized and the cause of application silos and perennial “are-we-there-yet=>No” answers to queries by top execs as to “Do I have information at my fingertips?”. Web Services , SOA, and ESB certainly help address these issues – but the core is being cross platform in code as well as data. And remember cross platform means talking and interacting between Win 98 and Win Xp or Linux and HP/UX.
Wow instead of getting insight I got incensed at the lightweight analysis of MDA. MDA has problems – but these are the least of them.