Sorry, Dion–Vietnam is still with us

Dion Almaer posted

    Ted Neward will be happy. It appears that the Vietnam that Ted thinks is ORM is coming to a close. In Gavin’s talk about Hibernate 3 at JavaPolis: There is talk that Hibernate 3 may be the last release. He was relaying the message that he feels that most of the problems involved in ORM have been taken care of, and that there are other problems to move on to. Is it the end of Vietnam? No more ORM innovation?

No, Dion, not quite. Stopping with Hibernate 3 simply means that Gavin’s unwilling to "put more boots on the ground", not that we’re out of Vietnam. How long do you think it will take some well-meaning-but-unaware-of-what-they’re-signing-up-for developer to volunteer to develop "Hibernate 4" in order to "just add one more thing"? That’s the crucial thing that makes this a quagmire–every time you start working with an O/RM utility/tool/framework, you find "just one more thing" that you’d like to add, and before too long, you’re wishing you could just carpet-bomb the whole thing and start over….

What we need (and even this doesn’t fix the problem entirely) is first-class tools (a la languages) that incorporate both concepts directly, not after-the-fact crutches that allow the underlying dissonance to peek through the layers of supposed abstraction. (And if you look at Hibernate, it doesn’t really abstract away all that much, to be honest; that’s partly why developers can be at least partially successful with it compared to those tools/frameworks/libraries that tried to completely hide away any notion of the RDBMS.)

Anybody interested in starting up an open-source SQL/J compiler?