What is the problem?

It seems that ray.util.multiprocessing.Pool.starmap does not work with iterable created from zip, unless it's explicitly converted to a list etc. For example, this one works:

from ray.util.multiprocessing import Pool as RayPool

with RayPool(4) as pool:
    print(pool.starmap(lambda x, y: x + y, list(zip(range(0, 7, 2), range(1, 8, 2)))))

But this one gives me an error, saying that TypeError: <lambda>() takes 2 positional arguments but 4 were given.

from ray.util.multiprocessing import Pool as RayPool

with RayPool(4) as pool:
    print(pool.starmap(lambda x, y: x + y, zip(range(0, 7, 2), range(1, 8, 2))))

With python's own multiprocessing.Pool I don't need to explicitly convert it to a list, so I think it'd be better if the behavior of ray.util.multiprocessing.Pool can be the same here? I guess this issue would be easy to fix.

Ray version and other system information (Python version, TensorFlow version, OS):

Ray 1.0.0 installed via pip, Python 3.6.11, SUSE 15.

Reproduction (REQUIRED)

Please provide a script that can be run to reproduce the issue. The script should have no external library dependencies (i.e., use fake or mock data / environments):

See above.

• I have verified my script runs in a clean environment and reproduces the issue.
• I have verified the issue also occurs with the latest wheels.