Hey hadoopers -

I've put together a python hdfs library and would love some feedback:

    http://github.com/traviscrawford/python-hdfs

Instead of wrapping the command-line ``hadoop`` tool like other python
hdfs libraries, this one uses ctypes to call libhdfs directly. It has
a file-like object for working with files, and a filesystem class for
interacting with non-file data.

I don't recommend using this in production yet, and not all methods
have been wrapped, but its usable and far enough along to demonstrate
this approach.

Thoughts/suggestions?

--travis