If you're doing something simple, I would recommend not going down that path until something absolutely requires a change. The biggest reason is that you will have a much larger build process. Essentially you'll need to compile all your python scripts into a jar, version them, then write a wrapper to launch the script (unless you want to explain java -jar awesome_program.jar from the command line). You'll also need to deal with JVM tuning on the client side (is the Heap large enough, garbage collection, etc.).
As long as your code is well documented, I don't think you'll have the problem you're thinking with client execution. Now, if you've got some crazy egg dependencies and don't think your users can handle the command line, that could change a packaging/deployment solution (though you can write a simple script to help out there).
Questions I would ask myself before proceeding:
- Does my code already work in Jython? If it does, use it!
If not:
- How much time is required to get up to speed on the new language differences?
- How much refactoring is involved?
- Does it make me more productive?
- Do I have time?
- Is the trade-off for "easier" desktop execution worth the extra effort in the build process (btw, you can automate most of this)
That said, I can say that I have experienced some weirdness in translating our Rails/Sinatra (in Ruby) apps to jRuby (though I was able to work through it pretty quickly). However, one of the coolest things I've been able to do is actually package up a Sinatra-based application as an executable jar file and write a quick batch file to start it up on a Windows desktop. I can't exactly speak to the Jython version of this, but the warbler gem, and the work the jruby folks have done, it's almost trivial to switch back and forth now. I would really hope (and almost assume) the Pythonists would have a similarly elegant solution to accomplish this (as long as Java is in the Windows classpath, which sometimes isn't the case).
