Field still incomplete

Over the decades, programmers have developed higher levels of abstraction in their languages. From the raw machine code of the 1940s to assembler to Fortran and later to C, C++, Java, C#. Once past assembler, all the higher constructs give us hardware independence. But suppose we have a system we want to code. The system could be defined textually or graphically. Currently, the implementation of a system is dependent on the software.

MDA is an attempt to give us software independence. The promise is to build a model according to some rules. This model documentation can then be run through a black box that makes implementation code. The user does not have to know the details of the black box. Exactly analogous to how you might write a C program and then turn it over to a compiler.

Clearly, there is immense value if MDA is possible. The authors say much of the value lies in the model being closer to what the user desires. Ideally, the user would draw up the model and be able to test it, without any knowledge of the lower, programming level. So there is no "verification gap". The document the user makes can be thus executed.

Some of you will remember similar claims made for fifth generation languages and their like back in the late 80s. These would turn every user into a programmer, eh?! Unfortunately, those efforts failed. The problem was too hard.

Sadly, as the authors themselves point out, MDA has not reached this goal either. A work still in progress. The book shows the current borders of research. It could do with some non-trivial examples. Important because of the abstract level of most of the discussion. Whilst there are some examples, they are of limited complexity.

Other Reviews

• An excellent introduction to MDA
• Field still incomplete