Compare:
>>> import pickle
>>> pickle.dumps(file('/etc/passwd'))
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File "/usr/lib/python2.4/pickle.py", line 1386, in dumps
    Pickler(file, protocol, bin).dump(obj)
  File "/usr/lib/python2.4/pickle.py", line 231, in dump
    self.save(obj)
  File "/usr/lib/python2.4/pickle.py", line 313, in save
    rv = reduce(self.proto)
  File "/usr/lib/python2.4/copy_reg.py", line 69, in
_reduce_ex
    raise TypeError, "can't pickle %s objects" %
base.__name__
TypeError: can't pickle file objects

with
>>> pickle.dumps(file('/etc/passwd'), protocol=2)
'\x80\x02c__builtin__\nfile\nq\x00)\x81q\x01.'

Similarly, the __reduce__ method works for basic
objects like str, int or dict with protocol=2, but
doesn't work with protocol=1:

>>> (1).__reduce_ex__(1)
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File "/usr/lib/python2.4/copy_reg.py", line 69, in
_reduce_ex
    raise TypeError, "can't pickle %s objects" %
base.__name__
TypeError: can't pickle int objects
>>> (1).__reduce_ex__(2)
(<function __newobj__ at 0xb7e5117c>, (<type 'int'>,
1), None, None, None)

That seems easy to fix by adding a dummy __reduce__ method to file. My
only worry is this could break file subclasses which may have ad-hoc
mechanisms implemented for pickling files.

Pickling a file is tricky. You can't just pickle the construction
parameters, you also need to save (and restore) the whole buffering
state. I'm not sure a half-working pickle solution would make a good
service to the user.