xudong963 commented on code in PR #20922:
URL: https://github.com/apache/datafusion/pull/20922#discussion_r2972685040
##########
datafusion/physical-plan/src/sorts/builder.rs:
##########
@@ -200,3 +229,143 @@ pub(crate) fn try_grow_reservation_to_at_least(
}
Ok(())
}
+
+/// Returns true if the error is an Arrow offset overflow.
+fn is_offset_overflow(e: &DataFusionError) -> bool {
+ matches!(
+ e,
+ DataFusionError::ArrowError(boxed, _)
+ if matches!(boxed.as_ref(), ArrowError::OffsetOverflowError(_))
+ )
+}
+
+fn offset_overflow_error() -> DataFusionError {
+ DataFusionError::ArrowError(Box::new(ArrowError::OffsetOverflowError(0)),
None)
+}
+
+fn recover_offset_overflow_from_panic<T, F>(f: F) -> Result<T>
Review Comment:
The panic-catching path is intentionally temporary — once
https://github.com/apache/arrow-rs/pull/9549 lands in DataFusion, `interleave`
will return a proper `OffsetOverflowError` instead of panicking, and we can
remove the `catch_unwind` + message matching entirely. At that point the
existing `retry_interleave` unit tests will cover the full production path
without any panic-payload coupling. Adding an expensive end-to-end test for a
short-lived workaround does not seem worth it.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
