ggangadharan commented on code in PR #6861:
URL: https://github.com/apache/hive/pull/6861#discussion_r4260242058
##########
ql/src/java/org/apache/hadoop/hive/ql/exec/vector/ptf/VectorPTFOperator.java:
##########
@@ -435,10 +435,11 @@ private void initBufferedColumns() {
* 1. order columns
* 2. the non-key input columns
* 3. any streaming columns that will already have their output values
- * 4. scratch columns (as they're used by input expressions of evaluators)
+ * 4. scratch columns holding the results of evaluator input expressions
*/
+ final int[] bufferedScratchIndices = getEvaluatorInputScratchIndices();
Review Comment:
Done. Removed the workaround from both files; they're now identical to their
versions before HIVE-25124. Both pass with PTF vectorization on, and their
.q.out files don't change.
Validated using below mvn test
```
mvn test -o -Pitests -pl itests/qtest -Dtest=TestMiniLlapLocalCliDriver
-Dqfile=sketches_rewrite_cume_dist.q,sketches_rewrite_cume_dist_partition_by.q
-Drat.skip=true
```
Validated below files as well, Seems these are Intentional non-vectorized
reference runs, so vectorized output can be compared against them
- vector_ptf_bounded_start
- count_distinct
- cume_dist
- lead_lag
- percent_rank
- windowing_over_const
- vector_windowing_row_number
--
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]
