Hyukjin Kwon created SPARK-61060:
------------------------------------
Summary: Vendor pyspark.conf and pyspark.taskcontext, and export
the names of pyspark, in the Python client as PySpark does
Key: SPARK-61060
URL: https://issues.apache.org/jira/browse/SPARK-61060
Project: Spark
Issue Type: Bug
Components: Connect
Affects Versions: connect-rust-4.3.0
Reporter: Hyukjin Kwon
In the PySpark client, {{pyspark}} exports the {{SparkConf}} of
{{pyspark.conf}}, the {{TaskContext}}, {{BarrierTaskContext}} and
{{BarrierTaskInfo}} of {{pyspark.taskcontext}}, {{Accumulator}},
{{AccumulatorParam}}, {{MarshalSerializer}}, {{CPickleSerializer}},
{{Profiler}}, {{BasicProfiler}} and {{keyword_only}}, and its {{since}} appends
the version to the docstring of a function. The modules {{pyspark.conf}} and
{{pyspark.taskcontext}} are pure Python.
The Rust-backed Python client has neither {{pyspark.conf}} nor
{{pyspark.taskcontext}}: {{from pyspark import TaskContext}} raises an
{{ImportError}}. Its {{SparkConf}} is a stub of {{set}}, {{get}}, {{remove}}
and a {{setAll}} of a dict, without the {{setAppName}}, {{setMaster}},
{{setIfMissing}}, {{contains}}, {{toDebugString}} and the other methods of
PySpark's, and {{pyspark}} lacks the other names. Vendor the modules and export
the names as PySpark does.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
