This is an automated email from the ASF dual-hosted git repository.
vgarg pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/hive.git
The following commit(s) were added to refs/heads/master by this push:
new 38574e3 HIVE-22164: Vectorized Limit operator returns wrong number of
results with offset (Ramesh Kumar Thangarajan, reviewed by Vineet Garg)
38574e3 is described below
commit 38574e3038dfdbf1c0b8b2c44bf4e94a281668d2
Author: Ramesh Kumar Thangarajan <rameshku...@cloudera.com>
AuthorDate: Mon Sep 9 22:02:54 2019 -0700
HIVE-22164: Vectorized Limit operator returns wrong number of results with
offset (Ramesh Kumar Thangarajan, reviewed by Vineet Garg)
---
itests/src/test/resources/testconfiguration.properties | 3 ++-
.../org/apache/hadoop/hive/ql/exec/vector/VectorLimitOperator.java | 4 +++-
.../results/clientpositive/llap/vector_windowing_order_null.q.out | 2 +-
3 files changed, 6 insertions(+), 3 deletions(-)
diff --git a/itests/src/test/resources/testconfiguration.properties
b/itests/src/test/resources/testconfiguration.properties
index 98280c5..34d0e27 100644
--- a/itests/src/test/resources/testconfiguration.properties
+++ b/itests/src/test/resources/testconfiguration.properties
@@ -425,7 +425,8 @@ minillap.query.files=acid_bucket_pruning.q,\
load_fs2.q,\
llap_stats.q,\
multi_count_distinct_null.q,\
- cttl.q
+ cttl.q,\
+ vector_offset_limit.q
minillaplocal.query.files=\
bucket_num_reducers_acid.q,\
diff --git
a/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorLimitOperator.java
b/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorLimitOperator.java
index 918a69a..79b073f 100644
--- a/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorLimitOperator.java
+++ b/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorLimitOperator.java
@@ -66,7 +66,9 @@ public class VectorLimitOperator extends LimitOperator
implements VectorizationO
public void process(Object row, int tag) throws HiveException {
VectorizedRowBatch batch = (VectorizedRowBatch) row;
- if (currCount + batch.size < offset) {
+ // We should skip number of rows equal to offset value
+ // skip until sum of current read count and current batch size less than
or equal offset value
+ if (currCount + batch.size <= offset) {
currCount += batch.size;
} else if (currCount >= offset + limit) {
setDone(true);
diff --git
a/ql/src/test/results/clientpositive/llap/vector_windowing_order_null.q.out
b/ql/src/test/results/clientpositive/llap/vector_windowing_order_null.q.out
index c6c39ed..e6a6a94 100644
--- a/ql/src/test/results/clientpositive/llap/vector_windowing_order_null.q.out
+++ b/ql/src/test/results/clientpositive/llap/vector_windowing_order_null.q.out
@@ -827,7 +827,6 @@ POSTHOOK: type: QUERY
POSTHOOK: Input: default@over10k_n21
#### A masked pattern was here ####
ts s sum_window_0
-2013-03-01 09:11:58.70307 calvin laertes 197097
2013-03-01 09:11:58.70307 calvin steinbeck 262874
2013-03-01 09:11:58.70307 david falkner 328506
2013-03-01 09:11:58.70307 fred nixon 394118
@@ -837,6 +836,7 @@ ts s sum_window_0
2013-03-01 09:11:58.70307 jessica laertes 656771
2013-03-01 09:11:58.70307 jessica polk 722558
2013-03-01 09:11:58.70307 katie king 788310
+2013-03-01 09:11:58.70307 katie white 853920
PREHOOK: query: explain vectorization detail
select s, i, round(sum(d) over (partition by s order by i desc nulls last) ,
3) from over10k_n21 limit 5
PREHOOK: type: QUERY
