[
https://issues.apache.org/jira/browse/ARROW-15018?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17532343#comment-17532343
]
Joris Van den Bossche commented on ARROW-15018:
-----------------------------------------------
This was fixed in ARROW-7914, and the index is now properly saved (if not a
RangeIndex) as columns in the feather file, and restored when converting back
to pandas.
> [Python] DataFrame Index modified during Feather serialization round trip
> -------------------------------------------------------------------------
>
> Key: ARROW-15018
> URL: https://issues.apache.org/jira/browse/ARROW-15018
> Project: Apache Arrow
> Issue Type: Bug
> Components: Python
> Affects Versions: 3.0.0, 6.0.1
> Reporter: Mike
> Priority: Major
>
> Some (or all) indexes are converted to a {{RangeIndex}} after a Feather
> serialization/deserialization round trip. I came across this in pyarrow 3.0.0
> and 6.0.1 for both a {{MultiIndex}} and an empty dataframe
> ({{{}Dataframe.index{}}} -> {{{}RangeIndex{}}}).
>
> Desired behavior: A dataframe with any type of index would be supported and a
> serialization/deserialization round trip would preserve the index and yield
> an identical dataframe
>
> Please feel free to modify priority and issue type as appropriate, thanks!
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
