The saga continues as Steve Kelly, from MetaCase, sent me an email, from which I'll quote the chunks relevant to the earlier Software Factories Discussion (here and here) on this blog. He also has his own blog post on this matter.
"I think it's important to separate the idea of Software Factories from their current MS implementation. Full code generation from Domain-Specific Modeling languages is something that has been done in practice for many years, e.g. David Weiss' work at Lucent to name just one. Tools for building these modeling languages, and giving them modeling tool support, have also existed for many years - including open source tools like GME and commercial products like MetaEdit+, both with over 10 years of real-world use.
The DSM Forum brings together all the various tool vendors (I'm a cofounder, and Microsoft signed up to join already at OOPSLA 2004). www.dsmforum.org contains lots of real world cases of DSM (or "Software Factories", if we're not fussy about terms). Maybe you could take a look, and perhaps point your readers there: looking at good examples of how something can be done successfully is generally more enlightening for newbies than looking how one team failed in their first project :-). (Perhaps MS should have picked a slightly smaller area than the whole healthcare industry for their first try!)
I've written some more comments on my blog, http://www.metacase.com/blogs/stevek/, about what's needed to make SF / DSM actually work. It's worth the effort - real world industrial cases consistently show 5-10x improvements in productivity. That's something the industry hasn't seen since the switch from assembler to 3GLs, and it's about time to raise the level of abstraction again!
Cheers,
Steve"
I like the way this conversation has evolved. This is serious discussion that I have no heard in any place that's "public" enough for real world people to read about. Personally, I'm learning more and more about this, and even though my view is slightly skewed from the Microsoft Perspective, it's important to realize there are other projects out there who seem to be doing just fine. I guess it all comes down to how big a chunk you want to bite off.
So "Is it possible" really becomes a "what scale is worth the effort". hm. Now there's a thought.
So "Is it possible" really becomes a "what scale is worth the effort". hm. Now there's a thought.