itholic commented on PR #41812: URL: https://github.com/apache/spark/pull/41812#issuecomment-1628534317
I apologize for the confusion. I will summarize the decision here to clarify the situation: By default, we should support all PySpark functionalities based on the latest version of pandas (currently 2.0.3), so this PR is appropriate change to reach our goal. So, also we don't need to test the CI with pandas 1.5.3, because we basically support the latest pandas. However, although we support PySpark based on the latest pandas, **we will not introduce any breaking changes from pandas 2.x in Apache Spark 3.5.0** since pandas 2.x introduces too many breaking changes so the users can get confuse if we introduce these all breaking changes in next minor release (Spark 3.5.0). For example, we should support new APIs or bug fixes introduced in pandas 2.x from Spark 3.5.0, but breaking changes such as API removals or behavior changes will be supported starting from Spark 4.0.0. The reason for aligning the CI with the latest version of pandas, even though we cannot immediately support all behaviors introduced in pandas 2.x, is that many users frequently install and use the latest versions of PySpark and pandas together. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
