diff options
Diffstat (limited to 'python/pyspark/rdd.py')
| -rw-r--r-- | python/pyspark/rdd.py | 13 |
1 files changed, 2 insertions, 11 deletions
diff --git a/python/pyspark/rdd.py b/python/pyspark/rdd.py index 5fb10f86f4..ed81eb16df 100644 --- a/python/pyspark/rdd.py +++ b/python/pyspark/rdd.py @@ -52,8 +52,6 @@ from pyspark.shuffle import Aggregator, ExternalMerger, \ get_used_memory, ExternalSorter, ExternalGroupBy from pyspark.traceback_utils import SCCallSiteSync -from py4j.java_collections import ListConverter, MapConverter - __all__ = ["RDD"] @@ -2317,16 +2315,9 @@ def _prepare_for_python_RDD(sc, command): # The broadcast will have same life cycle as created PythonRDD broadcast = sc.broadcast(pickled_command) pickled_command = ser.dumps(broadcast) - # There is a bug in py4j.java_gateway.JavaClass with auto_convert - # https://github.com/bartdag/py4j/issues/161 - # TODO: use auto_convert once py4j fix the bug - broadcast_vars = ListConverter().convert( - [x._jbroadcast for x in sc._pickled_broadcast_vars], - sc._gateway._gateway_client) + broadcast_vars = [x._jbroadcast for x in sc._pickled_broadcast_vars] sc._pickled_broadcast_vars.clear() - env = MapConverter().convert(sc.environment, sc._gateway._gateway_client) - includes = ListConverter().convert(sc._python_includes, sc._gateway._gateway_client) - return pickled_command, broadcast_vars, env, includes + return pickled_command, broadcast_vars, sc.environment, sc._python_includes def _wrap_function(sc, func, deserializer, serializer, profiler=None): |