language parallax

I thought I should at least make a small note of a discussion I had with Eric Van Wyk of the University of Minnesota during his recent visit to UCLA.

He’s probably the only person I’ve ever conversed with who had direct knowledge of what Intentional Software was up to — he was funded by them during his research at Oxford (actually I think they were still within Microsoft Research). (Other than a couple of folks from Intentional with whom I spoke at OOPSLA.)

He now casts a lot of that in terms of “attribute grammars”, which Don Knuth worked on as far back as the late 60’s. Though of course it sounds like the use of that phrase is somewhat controversial.

The languages discussed in his talk were java as the “host” language, sql, and a domain-specific language created for computational geometry. The main point emphasized is that any subset of languages described in his system can be composed together.

The most interesting point of discussion in our meeting was the value of “multi-view” programming, such as what intentional has. The other more concrete point is that it would be valuable for him (and probably the community at large) to have more examples of domain-specific languages. I mentioned to him a recent consulting experience I had with a company who had specifications that often exceeded 1000 pages long. With a domain-specific language, this could have been reduced by nearly a factor of 100.

That particular language would have looked something like Henzinger’s “Reactive Modules” language. I may try to create a business workflow language that looks something like RM next week as a part of this prospectus writing. If all goes according to plan, I’ll have code generation from this language to my Struts 2 / Hibernate 3 / AJAX / … framework.

With some extra added special sauce, of course. I’m settling on some new core abstractions that I’ll write more about as soon as I gain confidence that they will work.