Ahsen Jaffer's Blog

Investigating MDA part II November 08, 2005

In my opinion MDA has some real challenges, before it can gain widespread success.

The biggest problem I have is a philosophical one: MDA proposes that you can solve possibly any software engineering problem by starting from a high-level PIM (Platform Independent Model), and using complex transformations generate lower-level, and more detailed PSMs (Platform Specific Models), and eventually the running code.

Being a Java Developer I should be ashamed to agree with Jack Greenfield :) On a recent panel discussion (hosted here) he defended Microsoft's Software Factories vs. MDA. His biggest problem is similar to mine, "in the real world, we do not always know enough of a domain to model it effectively."

The other big problem: tons of PSMs will be required to be created and maintained. Every other month we see a new framework, or a new patch or an upgrade to the already abundant tools that we have out there. Who is going to take the responsibility of first of all creating all those PSMs and then maintaining them? The continuous requirement to update the models and the enormity of the effort itself might be a killer.

I am not saying it outright that MDA cannot be successfully at all. MDA can be a success, but under very strict situations: when the domain is very well known, and the correct PSMs exits. But as far as being an overwhelming success, I am sorry to acknowledge that it will most probably not happen, at least not in the near future.

Labels:




0 Comments:

Post a Comment

<< Home

back to top