Meetup uses it to some extent. I don't know how much, but I'm told by some of the programmers I know. I think it was Andrew who gave a Jython talk at Meetup HQ a few months back.

Jython is a great way to leverage the huge the amount of libraries and infrastructure available on the JVM. It also lets you reuse what you might already have, which might allow you to (covertly?) transition languages. I use it regularly.

If you're not already entrenched in the Java world, you probably don't need it. I use it because my clients are comfortable deploying WARs, EARs, JARs, SARs, etc. Oracle and IBM sell a full stack, so there's never any scrutiny.

Technologically, you might want it. Despite the tendency to ignore them, Java has a lot of superior tech that comes in handy. You'll get real threads, connection pooling, distributed transactions, fantastic async I/O libraries; Jython inherits all of this. If you need this, it's available. You don't have to roll your own. (You will have to learn it, though, and I think that's what really drives NIHS...) Understand that Jython does come with a performance penalty on some tasks, but pays dividends on others. You can also spot tune with pure Java to some extent.

So it's really a strategic thing. If you're selling into orgs that have established practices, or if you need really serious throughput, Jython will probably be a good choice. If you're building a web app that you run on your own infrastructure and you've already chosen Python and Django, you're probably better off without the extra layer.

Andrew (from Meetup) here. While the main web app is written in plain ole Java, the API is almost all Jython (the streaming stuff is Scala). The API team was 1 person until last December, and Jython was a big reason for this. The ability to glue all of the core logic and custom ORM into web.py on top of the modjy Servlet container is really powerful.

And, I concur entirely with what ibejoeb says.

Thanks for your great response. About the extra layer: http://news.ycombinator.com/item?id=2320299

I'm not sure about the development state of Jyhton, but like others have already said, using the JVM opens up an enormous box of Java platform goodies. With Jython you can build your web app front-end with Django and your backend business logic with Java, Scala or something else. It also gives you the ability to create worker thread etc.

This is a great online book about using Jython: http://www.jython.org/jythonbook/en/1.0

As of last week, Jython 2.5.2 was released, which is with some benchmarks is a 20% performance boost. I'm told that as of Pycon this past weekend, Jython 2.6 has been officially started, which will bring Jython closer to what you currently get with CPython 2.6 (library wise, etc). I can't speak for the project, but my guess is that it'll be sometime next year before that gets to an alpha state.

You might talk to some of the people at the Diamond Synchrotron facility in the UK. They are using it to allow scripting for control of their instrumentation. In my own use, I do remember there being some annoyances with getting a ^C into a text based client we were working on (from java), but that might have just been me. Also, I would suggest being careful with the possibility of memory leaks when calling java from jython (sorry, can't remember the details now).

> Jython 2.5.2 Has Been Released (March 3, 2011)

It's being actively developed. I know IBM uses it, at least in WebSphere.

About scalabilty I don't know, but it compiles to Java bytecode and the JVM is very mature (with Jython itself being mature as well), I'd wager that it scales well.

However, why are you using Jython with Django when you can use CPython? You're adding an additional layer of abstraction that will incur a performance hit.

Thanks for the response. Not CPython because we want something that runs on JVM and also be able to utilize existing Java code/technologies.