How significant is the introduction of annotations to the Java platform? How big a deal is that?

Well, [it's] sort of hard to quantify how big a deal annotations are, because in and of themselves they're actually not very useful because all you can do is add notes to things. And really what annotations are about is giving you a way to kind of hook between the language and meddling with the tools—or things that are bigger and beyond the language. And whether it's deployment tools or user interface generation tools or whatever, it's a way to build bridges between tools of various sorts and objects or methods or whatever. And I think there's incredible potential for really cool stuff to happen there.

But it's a little hard to say whether or not annotations are going to be the greatest thing ever, because it's so dependent on how people use them. There are plans right now, for instance, to change the way J2EE [Java 2 Platform, Enterprise Edition] deployment happens … to make the whole construction of deployment descriptors easier. I think that within the J2EE world that could be pretty revolutionary.

And I have a little back-burner project that … I've been working on that uses annotations as a way to build user interfaces for objects that I think is actually pretty cool, and I'm hoping to actually get it out there as an open-source project once it's not like totally embarrassing. But I think it could be very cool. So it's one of these things where I'm really pretty charged about it, but it all depends on what people do with it.

With NetBeans day occurring at JavaOne this year, I have to ask about Eclipse. Is the possibility of a Sun membership still on the table?

It's sort of hard to imagine us joining. I think that maybe Eclipse should join the NetBeans group.

So maybe you can cross-pollinate—no pun intended.

Well, the NetBeans open-source project has been up and running for four years now. It was up and running and running as an open-source project for over a year when IBM unveiled Eclipse. That whole part of it just felt really odd to me.

But when I wear my "church of Java" hat, tools are one of these things where it feels perfectly appropriate for there to be lots of them. And I think it's actually a good thing for there to be lots of them. The places where you need to have standards are in interfaces between things. That's where standards really pay off. And if you say there must only be one developer tool, all of a sudden you've killed innovation. It's a tragedy in the Windows world that there is only one developer tool.

You hear a lot of people say that if the Java camp doesn't get its act together, they're going to hand the developer space over to Microsoft. Is that even remotely possible or is it a given that diversity is the name of the game?

What do they mean by "get it together"? For me there's "get it together" in different flavors. There are things like standards around interfaces. And I think we do a pretty good job of that. Some of these are pretty chaotic, like all the J2EE stuff. But I think given all the temperatures and pressures of all the various actors involved in that space, I think it's astonishing that J2EE standards are in the shape that they're in when we've got ourselves and IBM and BEA and SAP and Oracle and a whole pile of people who really care about that stuff. And at some level every actor in the J2EE space wants to kill every other actor. And yet they all have this sort of uneasy coming together. And J2EE the standard is wonderful.

Where I have a problem with the whole "get your act together" phrase is that innovation depends on ideas and things mixing and somebody having the idea to go off and do something neat. If somebody can't go off and invent something cool—a developer tool or whatever—the world is a worse place.

I think the world got to a really tragic point when there was essentially one word processor in the entire universe. You can't tell me that for the rest of time all innovation in document production is over because Microsoft Word is the end of it. And in a sense, it had become the end of it because if you had gone to any funding source and said, "Hi, I'd like to write a word processor for profit," if you'd gone to any VC in Silicon Valley and said, "I want to compete against Microsoft Word," you would have become the stuff of legend at cocktail parties with people laughing so hard. It's such a stupid business plan.

And yet there were these nut cases in Hamburg who built this thing and they couldn't really make it fly. So we bought them and helped them a little and pretty soon it turned into Star Office. So I think it's a tremendous disservice to the universe to have only one. The place where you need to have standards is around the interfaces. And if we "got our act together" we'd be dead.

To be in the Java world you'd have to believe in Darwinism. At some level it's chaos, and at some level it's innovation, it's cleverness and people trying to find a better way. And I guess I'm a fan of evolution.

Next page: Open-sourcing Java.