Writing Node.js modules in ClojureScript
blog.sourceninja.com
blog.sourceninja.com
I think the argument for Node.js was that it is typically difficult to write asynchronous code, and they made some effort to make that better, and the one language across front end and back.
With Clojure you can write asynchronous code just as easily with much better performance. I can also share code bases across my backend (cascalog, storm), mid-stack (noir), and front-end ClojureScript.
The arguments against Clojure is that some people have difficulty learning it and functional programming in general.
That said, my argument against Clojure is the sheer weight Java interop demands (I always found myself dealing with far more Java than I would have liked). And with Clojurescript, the tooling feels incredibly raw to me. Getting started sucked up most of a weekend, and even then it was considerably less pleasant than parenscript.
That said, clojurescript really just felt very young to me, and could become something amazing with time and momentum. Perhaps you could write a short article about your setup (backend-mid-frontend), and some tips for people testing the waters? Would be highly appreciated!
The high level is Rails on the very front for legacy reasons, and it's easier to find libraries that do, authentication, authorization, billing etc... that speaks to solr or a restful service running on noir hosted by tomcat, cassandra is the main DB. All backend tasks are performed by Hadoop or storm, and those jobs are all written in Clojure.
To be honest I feel that the Java interop isn't that bad. It's easier than doing interop from ruby/python down to C or C++. It provides access to lots of high quality libraries and you get the JVM. I think it's more than worth the price.
Although I know many functional programming languages, Clojure is the first that I've stuck with long term, and I think the strength of the ecosystem has a lot to do with it. Languages like Erlang, Haskel, and lisp are really interesting, but the breadth and depth of the libraries isn't comprable. Being a small company we need to be careful how we allocate resources. We contribute back where we can, for instance we contribute to clj-hector and cassandra.cascading but we're only a two man dev shop.