Hyukjin Kwon created SPARK-61108:
------------------------------------
Summary: Name the nested fields of collected rows by the schema,
and raise for calendar intervals in collect, in the Python client as PySpark
does
Key: SPARK-61108
URL: https://issues.apache.org/jira/browse/SPARK-61108
Project: Spark
Issue Type: Bug
Components: Connect
Affects Versions: connect-rust-4.3.0
Reporter: Hyukjin Kwon
In PySpark Connect, {{DataFrame.collect}} (and so {{first}}, {{head}}, {{take}}
and {{tail}}) checks the Arrow schema of the result with {{from_arrow_schema}}.
It then builds the rows by the schema of the result:
* The fields of nested structs have the names of the schema. Duplicate names
are kept: {{SELECT named_struct('a', 1, 'a', 2) AS v}} collects
{{Row(v=Row(a=1, a=2))}}, while the server's Arrow data names them {{a_0}},
{{a_1}}. {{toLocalIterator}} names them the same way.
* A calendar interval raises {{UNSUPPORTED_DATA_TYPE_FOR_ARROW_CONVERSION}}
("month_day_nano_interval is not supported in conversion to Arrow").
The Python client instead named nested fields by the Arrow data:
{{Row(v=Row(a_0=1, a_1=2))}}, where {{v.a}} raised {{ATTRIBUTE_NOT_SUPPORTED}}.
It also collected a calendar interval as a string such as "14 months 25 days
18367500000000 ns".
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
