deniskuzZ commented on code in PR #5868:
URL: https://github.com/apache/hive/pull/5868#discussion_r2165127166
##########
ql/src/java/org/apache/hadoop/hive/ql/io/BatchToRowReader.java:
##########
@@ -518,7 +521,8 @@ public static TimestampWritableV2
nextTimestamp(ColumnVector vector,
result = (TimestampWritableV2) previous;
}
TimestampColumnVector tcv = (TimestampColumnVector) vector;
- result.setInternal(tcv.time[row], tcv.nanos[row]);
+ result.set(Timestamp.ofEpochSecond(Math.floorDiv(tcv.time[row], 1000L),
tcv.nanos[row],
+ tcv.isUTC() ? ZoneOffset.UTC : ZoneId.systemDefault()));
Review Comment:
`TreeReaderFactory` is using `ReaderContext` to propagate the flag into the
`TimestampColumnVector` vector.
````
result.setIsUTC(this.context.getUseUTCTimestamp());
````
Rather than passing the context to the vector, we loop through all column
vectors, searching for `TimestampColumnVector` and setting the flag manually in
the RecordReaderImpl constructor.
While I agree that apache/orc#2300 is a proper fix, can we do the refactor
in the meantime.
Additionally, it would be nice to reuse the timestamp conversion function to
avoid code duplication
--
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]
