Robert Clarisó sent me this excerpt form the Hype Cycle de Gartner for Application Development 2014 (link, but you’ll need to have a Gartner account).
Model-driven architectures (MDA) became obsolete before reaching full maturity. This a reflection of the market shift to more general, model-driven development approaches. Application platform as a service (aPaaS), business process management suites (BPMSs) and high-productivity platforms all share heritage in development by abstraction and rule and metadata specification. Pure-play MDA providers will continue to be tracked, and Gartner will continue to cover MDA. In the future, coverage will be focused on model-driven engineering in all its forms, not just MDA.
To be honest, this is hardly a surprise. When it first appear in 2003 (MDA Guide v1), MDA was a good marketing/teaching concept to introduce people to the world of model-driven engineering but it quickly became clear that to address the variety of market needs we needed a more flexible concept and it ended up fading away (I bet very few people know that in fact, the OMG “recently” published a second version of the MDA guide). In defense of MDA, I have to say that MDA in itself was quite flexible already but most people took it in its most restrictive view (one PIM – on PSM – final code process) and I guess that after a while, tool vendors had to come up with alternative names to keep selling their tools so its fate was sealed anyway.
So, let’s all chant: “MDA is dead, long live MDE” (and if you’re still not sure about the differences between them, check this page clarifying concepts: MBE vs MDE vs MDD vs MDA ).
Featured image by Kev-shine
FNR Pearl Chair. Head of the Software Engineering RDI Unit at LIST. Affiliate Professor at University of Luxembourg. More about me.
To me, MDA created an enormous expectation that industry couldn’t (wanted?) to fill. Seemed to me it was expected each vendor should build their full toolchain, and I never appreciated significant interoperability between then. More important: the lack of shareable models in, preferably, open repositories detracted value and interest to model driven approach.
I know I have an strong OSS bias but I’m sure a commitment on public models repos (mainly CIMs & PIMs) would have lead to a giant leap, but I guess this kind of great effort on standardization is not in the industry agenda.
About the repos part, I totally agree and we have tried but without success so far: https://modeling-languages.com/one-virtual-model-repository-rule/