Github user HyukjinKwon commented on the issue:
    Ah, nope. If PyArrow version is lower then we claim, for example, 0.7.0, 
seems tests go failed:
    ERROR: test_vectorized_udf_wrong_return_type 
    Traceback (most recent call last):
      File "/.../spark/python/pyspark/sql/", line 4019, in 
        f = pandas_udf(lambda x: x * 1.0, MapType(LongType(), LongType()))
      File "/.../spark/python/pyspark/sql/", line 2309, in 
        return _create_udf(f=f, returnType=return_type, evalType=eval_type)
      File "/.../spark/python/pyspark/sql/", line 47, in _create_udf
      File "/.../spark/python/pyspark/sql/", line 132, in 
        "however, your version was %s." % pyarrow.__version__)
    ImportError: pyarrow >= 0.8.0 must be installed on calling Python process; 
however, your version was 0.7.0.
    Ran 33 tests in 8.098s
    FAILED (errors=33)
    Will clarify it in PR description.


To unsubscribe, e-mail:
For additional commands, e-mail:

Reply via email to