James Turton created DRILL-8403:
-----------------------------------
Summary: Rewritten aggregate functions are incorrectly grouped
when used with PIVOT
Key: DRILL-8403
URL: https://issues.apache.org/jira/browse/DRILL-8403
Project: Apache Drill
Issue Type: Bug
Affects Versions: 1.21.0
Reporter: James Turton
Assignee: Vova Vysotskyi
Fix For: 1.21.1
The following query should group aggregates by both marital_status and
education_level but only groups them by education_level.
apache drill> SELECT
2..semicolon> *
3..semicolon> FROM
4..semicolon> (SELECT
5..........)> education_level,
6..........)> salary,
7..........)> marital_status,
8..........)> extract(year from age(birth_date)) age
9..........)> FROM
10.........)> cp.`employee.json`)
11.semicolon> PIVOT (
12.........)> avg(salary) avg_salary, avg(age) avg_age FOR marital_status IN
('M' married, 'S' single)
13.........)> );
+---------------------+--------------------+--------------------+--------------------+--------------------+
| education_level | married_avg_salary | married_avg_age |
single_avg_salary | single_avg_age |
+---------------------+--------------------+--------------------+--------------------+--------------------+
| Graduate Degree | 4392.823529411765 | 100.32352941176471 |
4392.823529411765 | 100.32352941176471 |
| Bachelors Degree | 4492.404181184669 | 102.22996515679442 |
4492.404181184669 | 102.22996515679442 |
| Partial College | 4047.1180555555557 | 100.10069444444444 |
4047.1180555555557 | 100.10069444444444 |
| High School Degree | 3516.1565836298932 | 103.12811387900356 |
3516.1565836298932 | 103.12811387900356 |
| Partial High School | 3511.0852713178297 | 102.30232558139535 |
3511.0852713178297 | 102.30232558139535 |
+---------------------+--------------------+--------------------+--------------------+--------------------+
5 rows selected (0.285 seconds)
00-00 Screen : rowType = RecordType(ANY education_level, ANY
married_min_salary, DOUBLE married_avg_age, ANY single_min_salary, DOUBLE
single_avg_age): rowcount = 46.3, cumulative cost = \{1486.23 rows,
35748.229999999996 cpu, 474630.0 io, 0.0 network, 8148.800000000001 memory}, id
= 812
00-01 Project(education_level=[$0], married_min_salary=[$1],
married_avg_age=[$2], single_min_salary=[$3], single_avg_age=[$4]) : rowType =
RecordType(ANY education_level, ANY married_min_salary, DOUBLE married_avg_age,
ANY single_min_salary, DOUBLE single_avg_age): rowcount = 46.3, cumulative cost
= \{1481.6 rows, 35743.6 cpu, 474630.0 io, 0.0 network, 8148.800000000001
memory}, id = 811
00-02 Project(education_level=[$0],
married_min_salary=[divide(CastHigh(CASE(=($2, 0), null:NULL, $1)), $2)],
married_avg_age=[divide(CastHigh(CASE(=($4, 0), null:NULL, $3)), $4)],
single_min_salary=[divide(CastHigh(CASE(=($2, 0), null:NULL, $1)), $2)],
single_avg_age=[divide(CastHigh(CASE(=($4, 0), null:NULL, $3)), $4)]) : rowType
= RecordType(ANY education_level, ANY married_min_salary, DOUBLE
married_avg_age, ANY single_min_salary, DOUBLE single_avg_age): rowcount =
46.3, cumulative cost = \{1435.3 rows, 35512.1 cpu, 474630.0 io, 0.0 network,
8148.800000000001 memory}, id = 808
00-03 HashAgg(group=[\{0}], agg#0=[$SUM0($2)], agg#1=[COUNT($2)],
agg#2=[$SUM0($3)], agg#3=[COUNT($3)]) : rowType = RecordType(ANY
education_level, ANY $f1, BIGINT $f2, BIGINT $f3, BIGINT $f4): rowcount = 46.3,
cumulative cost = \{1389.0 rows, 34725.0 cpu, 474630.0 io, 0.0 network,
8148.800000000001 memory}, id = 807
00-04 Project(education_level=[$0], marital_status=[$1],
salary=[$2], age=[EXTRACT(FLAG(YEAR), AGE($3))], $f4=[IS TRUE(=($1, 'M'))],
$f5=[IS TRUE(=($1, 'S'))]) : rowType = RecordType(ANY education_level, ANY
marital_status, ANY salary, BIGINT age, BOOLEAN $f4, BOOLEAN $f5): rowcount =
463.0, cumulative cost = \{926.0 rows, 8797.0 cpu, 474630.0 io, 0.0 network,
0.0 memory}, id = 806
00-05 Scan(table=[[cp, employee.json]], groupscan=[EasyGroupScan
[selectionRoot=classpath:/employee.json, numFiles=1,
columns=[`education_level`, `marital_status`, `salary`, `birth_date`],
files=[classpath:/employee.json], usedMetastore=false, limit=-1,
formatConfig=JSONFormatConfig [extensions=[json]]]]) : rowType = RecordType(ANY
education_level, ANY marital_status, ANY salary, ANY birth_date): rowcount =
463.0, cumulative cost = \{463.0 rows, 1852.0 cpu, 474630.0 io, 0.0 network,
0.0 memory}, id = 805
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
