Hyukjin Kwon created SPARK-61154:
------------------------------------
Summary: Raise VALUE_NOT_BETWEEN for a literal of an int beyond
the range of a long, in the Python client as PySpark does
Key: SPARK-61154
URL: https://issues.apache.org/jira/browse/SPARK-61154
Project: Spark
Issue Type: Bug
Components: Connect
Affects Versions: connect-rust-4.3.0
Reporter: Hyukjin Kwon
A Python int outside the int64 range became a double in the Python drop-in, so
comparisons were silently wrong: df.where(F.col("a") == 2**63) returned the
rows 9223372036854775807 and 9223372036854775806, and lit(2**70) was a double.
PySpark's LiteralExpression._infer_type raises PySparkValueError
VALUE_NOT_BETWEEN (arg_name "value", min -9223372036854775808, max 32767) for
such an int, and fillna raises ValueError "Value out of range" for it.
The fix applies to every literal path of the drop-in (lit, Column operators,
isin, between, when, arrays, pivot values); createDataFrame already matched
PySpark (OverflowError).
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
