Yes, compojure is early, but for us it has been totally worthwhile. A lisp that has 1) a fast implementation 2) excellent support for parallelization and 3) well supported web servers has been a huge win. We have had to reinvent a few wheels, but I would much rather re-invent e.g. user logins than Clojure + the JIT, debugger & profiler support of the JVM.
Clojure is going to get better webapp support, Ruby isn't going to get better at number crunching or parallelization or get rid of its warts. So yes, we've been very happy.
Is Clojure good at that? Or do you use Java for actual heavy lifting?
Why not Python with its many numerical libraries (numpy, scipy, etc.) ?
The power of using rails comes, partly, from all the great really simple to use plugins and libraries. A lot of those libraries have C dependencies, because of that using jruby on rails is poor solution imho. I think C ruby is a viable option though.
Compojure is mature enough to take on those frameworks. In comparison with grails and groovy, clojure is simple enough to work as expected while groovy has some excentricities.
Building a web application in Clojure right now seems to involve a large amount of re-inventing the wheel. Meanwhile, we can just switch over to Rails and have stuff working in minutes.
Does anyone have any personal anecdotal evidence that Clojure's benefits outweigh this chief drawback at this time? What are the chances that Clojure will ever be where Ruby (with Rails) is now?