caneff commented on PR #28523: URL: https://github.com/apache/beam/pull/28523#issuecomment-1727496476
> Are some tests failing on Pandas2 if we don't set it `copy=True`? > > I think I understand the idea of preserving the behavior but also concerned since this will be a difference in defaults b/w Beam and Pandas. > > Also, we could call this out in https://beam.apache.org/documentation/dsls/dataframes/differences-from-pandas/ . Looking at it again, we don't support copy=False anywhere in Beam. Because it requires memory sharing semantics we can't support. I don't know if that's worth calling out now because it is no different than ever before. If you ever specify copy=False currently it will either raise an error or ignore it and copy anyway depending on the function. Therefore the current PR behavior of setting the default to True makes sense because False never will. It is just before we relied on the default in pandas being True where it is now None. -- 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]
