Skip to main content

Ken Arnold Critiques Ant

Posted by duncan on June 18, 2003 at 8:54 PM PDT

I just ran into Ken Arnold at MacHack (he's giving the keynote tonight at midnight) and after chatting a bit, we picked back up the Ant conversation we were having at JavaOne. He's obviously been thinking about it for a while because he came out with the line:

The problem with Ant is that it violates something we learned with Unix. Tasks aren't composable.

And you know, he's right. Ant uses the file system to store state between tasks. There's no inherit way to pass the results of one task to another. Sure, you can set variables and what not, but... Ah well. The way Ant is today is pretty much where it's going to be for a while. Lots of people are using it and inertia is powerful indeed. Look at make for an example of what inertia will do. (Disclaimer: I'm not working directly on Ant anymore. That's the domain of the Apache Ant Project)

The cool thing about having your ideas out there in public source code is that everybody can help improve it. The odd thing is that everybody can critique design decisions that you made a long long long time ago. But that's actually cool in my book. How often do you get an uber-hacker like Ken critiquing something that you did? In a friendly, thoughtful, and well argued way? In any case, I think Ken is trying to rekindle my interest in building tools again. You never know...