lidavidm commented on a change in pull request #29818:
URL: https://github.com/apache/spark/pull/29818#discussion_r506588271
##########
File path: python/pyspark/sql/pandas/conversion.py
##########
@@ -103,10 +103,22 @@ def toPandas(self):
batches = self.toDF(*tmp_column_names)._collect_as_arrow()
if len(batches) > 0:
table = pyarrow.Table.from_batches(batches)
+ del batches
# Pandas DataFrame created from PyArrow uses
datetime64[ns] for date type
# values, but we should use datetime.date to match the
behavior with when
# Arrow optimization is disabled.
- pdf = table.to_pandas(date_as_object=True)
+ pandas_options = {'date_as_object': True}
+ if selfDestruct:
+ # Configure PyArrow to use as little memory as
possible:
+ # self_destruct - free columns as they are
converted
+ # split_blocks - create a separate Pandas block
for each column
+ # use_threads - convert one column at a time
+ pandas_options.update({
+ 'self_destruct': True,
+ 'split_blocks': True,
Review comment:
Not quite _necessary_ but good to have, else we may end up allocating a
large block if there are a lot of columns of the same type, defeating the
point. Pandas may reconsolidate later but that's part of the issue of using
Pandas.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
