[ARROW-4867] [Python] Table.from_pandas() column order not respected

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 0.12.1
• Fix Version/s: 0.13.0
• Component/s: Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/21377
• Assignee: Wes McKinney
• Reporter: Andreas
• Created: 14/Mar/19 12:37
• Updated: 11/Jan/23 07:36
• Resolved: 16/Mar/19 16:33

Description

Starting with 0.12, the order of the columns in Table.from_pandas is no longer identical to the order of the colums given in the columns argument. This used to be the case in previous versions.

0.12:

In [1]: pyarrow.Table.from_pandas(pd.DataFrame(dict(a=[1],b=[2])), columns=['b', 'a'])
Out[1]:
pyarrow.Table
a: int64
b: int64

0.11:

In [1]: pyarrow.Table.from_pandas(pd.DataFrame(dict(a=[1],b=[2])), columns=['b', 'a'])
Out[1]:
pyarrow.Table
b: int64
a: int64

Issue Links

• links to: GitHub Pull Request #3930