JBoss 4 and AOP, what gives with it?

Is JBoss 4.0DR1 purely a JavaOne release? The empty change log for the release at SF is the best immediate inidication that this is the case, but a little context might drive the point home.

Aspect-Oriented Programming is brilliant and JBoss' commitment to pushing it forward is admirable. But is it really necessary at this stage in the game? Is it possible that The JBoss Group could be so forward thinking that they're leaving us all stranded in the present? Let's make this clear--we understand the vision, but we're having a little bit of trouble turning it into reality.

Right now most companies are just getting comfortable with servlets and JSP and hooking straight into the database with JDBC. The really progressive companies are getting started on real EJB solutions, and JBoss definitely can help them out in this arena. Ahead of them are the advanced EJB/CMR and XDocleteers who are pushing the edges of EJB technology and helping to define the future of the spec. I consider those guys bleeding edge - they could turn around at any moment and redefine everything we take for granted. One step beyond them are the web services teams bringing 30 year old legacies to the web and unlocking real industrial strength puzzles. We'll call those guys the whistling edge. Finally, off in the ultra-abstract world of Aspect-Oriented Programming is JBoss. And I'm wondering if that gives them an edge at all.

AOP and the wonders of instrumentation won't help us this year. Instrumentation is just another way of saying "rewrite the compiled code as its loaded into memory". I don't know about your CIO, but I'd be pretty sketchy about letting anyone do this with my J2EE app anytime in the near future. Way down here in userland our idea of instrumentation is better debug messages. Or maybe a management console that can save properties between server reboots. How about some runtime monitoring tools? We see their perspective just fine. Do they see ours?

There was a promised complete rewrite of the JMS implementation, which has not been touched in this release. The promised rewrite of the persistence engine is also not present. To be fair, things have improved considerably in version 3.2, but the management features deserve some of the focus that AOP is getting.

Increased separation of concerns bringing us the power of EJB without all of the development overhead and ridiculous complexity is the promise of AOP. This would be a great move forward. But are we ready for it yet? Do we even need it yet? Can Java as a language even handle it? These are some of the hard questions we'll be asking at JBossTwo later this week. µ