Stevo Mitric created SPARK-58737:
------------------------------------
Summary: Support nanosecond-precision timestamps in
ColumnarRow/ColumnarBatchRow copy() and get()
Key: SPARK-58737
URL: https://issues.apache.org/jira/browse/SPARK-58737
Project: Spark
Issue Type: Task
Components: Spark Core
Affects Versions: 4.4.0
Reporter: Stevo Mitric
The generic type-dispatch in ColumnarRow and ColumnarBatchRow
(sql/catalyst/.../vectorized/) does not handle the nanosecond-precision
timestamp types TimestampNTZNanosType(p) / TimestampLTZNanosType(p) (p in [7,
9], carried as org.apache.spark.unsafe.types.TimestampNanosVal).
As a result, copy() and get(int, DataType) throw at runtime on a nanosecond
timestamp column:
- copy() dispatches on PhysicalDataType and falls through to
RuntimeException("Not implemented. " + dt).
- get(int, DataType) dispatches on DataType and throws
SparkUnsupportedOperationException (_LEGACY_ERROR_TEMP_3155 in ColumnarRow,
_LEGACY_ERROR_TEMP_3152 in ColumnarBatchRow).
The per-cell leaf accessors (getTimestampNTZNanos / getTimestampLTZNanos)
already exist and work; only these two generic dispatch methods were never
extended.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
