[
https://issues.apache.org/jira/browse/BEAM-11929?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17309050#comment-17309050
]
Robert Bradshaw edited comment on BEAM-11929 at 3/25/21, 11:52 PM:
-------------------------------------------------------------------
Per the recent thread on the dev list, this may be more severe.
{code:python}
with beam.Pipeline() as p:
data = p | beam.Create([None]) | beam.Map(lambda x: beam.Row(c='c', b='b',
a='a'))
res = to_pcollection(to_dataframe(data))
res | beam.Map(print)
# BeamSchema_faf9f167_bf18_43b2_b091_a6434aba66fd(c='a', b='b', a='c')
{code}
was (Author: robertwb):
Per the recent thread on the dev list, this may be more severe.
with beam.Pipeline() as p:
data = p | beam.Create([None]) | beam.Map(lambda x: beam.Row(c='c', b='b',
a='a'))
res = to_pcollection(to_dataframe(data))
res | beam.Map(print)
> DataframeTransfom, BatchRowsAsDataFrame do not preserve field order when
> schema created with beam.Row
> -----------------------------------------------------------------------------------------------------
>
> Key: BEAM-11929
> URL: https://issues.apache.org/jira/browse/BEAM-11929
> Project: Beam
> Issue Type: Bug
> Components: sdk-py-core
> Affects Versions: 2.26.0, 2.27.0, 2.28.0, 2.29.0
> Reporter: Brian Hulette
> Assignee: Brian Hulette
> Priority: P2
> Labels: dataframe-api
> Fix For: 2.30.0
>
> Time Spent: 5h
> Remaining Estimate: 0h
>
> The workaround is to use a NamedTuple instance with DataframeTransform.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
