testing arbitrary backend implementations dynamically using py.test parametrized tests

Question

trying to write a bunch of tests for a program that supports loading an arbitrary backend implementation dynamically using py.test parametrized tests, and it was a bit more complicated than I expected. What I'm intending to do (and not completely achieving) is to use the list_backends method below to parametrize some tests based on the number of backend implementations that exist at runtime. Something similar to my current implementation is below, where I copy a bunch of config files, one for each backend, to tmpdir and then basically return a particular section of the config file named after the backend:

  @pytest.fixture(scope='function')
  def backend(request):
      """Return the backend implementation class."""
      return get_backend(request.param)


  @pytest.fixture(scope='function')
  def configdir(tmpdir, request):
      """Copy all test config files to a temporary directory."""
      testdir = py.path.local(request.module.__file__).dirname
      src = testdir.join("configs")
      dest = tmpdir.join("configs").mkdir()
      for f in src.visit():
          f.copy(dest)
      return dest


  @pytest.fixture(scope='function')
  def config(configdir, request):
      """Return the appropriate config section for this test."""
      cfgfile = configdir.join(request.param)
      cfg = read_config(cfgfile)
      return cfg[request.param]


  @pytest.mark.parametrize(
S>    'backend,config', zip(*[list_backends()]*2), indirect=['backend', 'config'])
  def test_backend_with_config(backend, config):
      """Run some tests with a particular backend and config."""
      instance = backend(config)
      # test stuff

  ...

First, am I overdoing this? Second, is there a simpler way to do this kind of parameterization with py.test?