Running on alternative Python implementations

Pyro is written in 100% pure Python which -theoretically- enables it to be used with any compatible Python implementation. There are a few gotchas however.

Note

If possible please use the most recent version available of the Python implementation.

Jython

Jython

Jython is a Python implementation running on the Java virtual machine.

  • The multiplexing server type (select/poll-based server) is not reliable on Jython. You can only use the threadpool server type.
  • You cannot use the other parameter to the requestloop of a Threadpool server. The workaround is to spawn a separate thread for each socket that you need to listen to. (The name server does this for the broadcast server, if it detects that it is running on Jython)

IronPython

IronPython

IronPython is a Python implementation running on the .NET virtual machine.

  • Pyro requires the zlib module, which is not included in older IronPython versions. IronPython 2.7 includes it by default. If you need to install it manually, get it from the developer.
  • IronPython cannot properly serialize exception objects, which could lead to problems when dealing with Pyro’s enhanced tracebacks. For now, Pyro contains a workaround for this IronPython bug.

Pypy

Pypy

Pypy is a Python implementation written in Python itself, and it usually is quite a lot faster than the default implementation because it has a JIT-compiler.

I haven’t used Pypy much let alone with Pyro, but it seems that at least the recent Mac OS X and Linux builds of Pypy work fine with Pyro. There are a lot of problems on Windows however which I can’t explain (most likely due to the fact that the windows build of Pypy is quite old and misses a lot of bug fixes). So don’t use Pypy on Windows with Pyro.

There is one problem with running the unit tests on Pypy: one of the tests freezes and Pypy cannot complete the test suite. I hope to fix this in the near future (or maybe it’s a bug in Pypy?)