The Future of Model Driven Development

The Model Driven Architecture (MDA) initiative started with very high objectives, but suffered a severe setback by prematurely declaring and marketing "industry standards" that did not live up to users' expectations. The resulting problems were already identified back in 2002 ("Will MDA Still Be Around In Five Years Time?"), and have led to the Open Source Generative Modeling Technologies project that now features several valuable tool components for model driven software development.

The following comments are from people who have been directly involved in MDA standardization initiatives, and from people who are integrating and interfacing MDA-compliant tools into end-to-end tool chains for software production:

The UML/MOF/QVT etc standards process was significantly hampered by the lack of a reference implementation. I suspect that this applies to other standards too: where a reference implementation is used as part of the standardization process the resulting standard is *much* more useful. In many ways, the (lack of) RI issue is solved completely by taking an open-source approach because the process tends to start out with an implementation and then natural selection takes over. A standard must fit into a process and therefore *do* something. I make this point because I think that there is nothing inherently wrong about standardization, however the basis for many software standardization processes is seriously flawed.

The real problem is that standards constructed by consortia way have committee-construction problems, and that produces camels, not fighter jets.

It is time to move on, and to use the KISS initiative to map out a more realistic path towards interoperability.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License