mirror of
https://github.com/apache/impala.git
synced 2026-01-01 18:00:30 -05:00
63f17e9cea
Previously, scanners will assume that there are no conjuncts associated with a scan node for queries with no materialized slots (e.g. count(*)). This is not necessarily the case as one can write queries such as select count(*) from tpch.lineitem where rand() * 10 < 0; or select count(*) from tpch.lineitem where rand() > <a partition column>. In which case, the conjuncts should still be evaluated once per row. This change fixes the problem in the short-circuit handling logic for count(*) to evaluate the conjuncts once per row and only commits a row to the output row batch if the conjuncts evaluate to true. Testing done: Added the example above to the scanner test Change-Id: Ib530f1fdcd2c6de699977db163b3f6eb38481517 Reviewed-on: http://gerrit.cloudera.org:8080/8623 Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com> Reviewed-by: Alex Behm <alex.behm@cloudera.com> Reviewed-by: Dan Hecht <dhecht@cloudera.com> Tested-by: Impala Public Jenkins
101 lines
2.5 KiB
Plaintext
101 lines
2.5 KiB
Plaintext
====
|
|
---- QUERY
|
|
# This query will do a full table scan, doing a simple aggregation on all cols with
|
|
# a simple predicate
|
|
select count(*),
|
|
sum(id), count(bool_col), sum(tinyint_col), sum(smallint_col),
|
|
sum(int_col), sum(bigint_col), max(float_col), max(double_col),
|
|
max(date_string_col), max(string_col), max(timestamp_col)
|
|
from alltypesagg
|
|
where id % 2 = 0 and day is not null
|
|
---- RESULTS
|
|
5000,24995000,5000,20000,245000,2495000,24950000,1097.800048828125,10079.8,'01/10/10','998',2010-01-10 18:00:55.300000000
|
|
---- TYPES
|
|
BIGINT, BIGINT, BIGINT, BIGINT, BIGINT, BIGINT, BIGINT, FLOAT, DOUBLE, STRING, STRING, TIMESTAMP
|
|
====
|
|
---- QUERY
|
|
# This query will do a join, projecting one string col from each table.
|
|
# This is interesting because the join contains string cols which causes the scanners
|
|
# to do different memory handling.
|
|
select sum(t1.id), sum(t1.int_col),max(t1.date_string_col), max(t2.string_col)
|
|
from alltypesagg t1
|
|
inner join alltypesagg t2
|
|
on t1.id = t2.id and t1.day is not null and t2.day is not null
|
|
---- RESULTS
|
|
49995000,4995000,'01/10/10','999'
|
|
---- TYPES
|
|
BIGINT, BIGINT, STRING, STRING
|
|
====
|
|
---- QUERY
|
|
# This query does a top-n on non-string cols. This is different because without
|
|
# string cols, scanners will handle io buffers differently. They don't need to
|
|
# be passed up the execution tree.
|
|
select id, bool_col, int_col
|
|
from alltypesagg where day is not null
|
|
order by 1 desc, 2 desc, 3 desc
|
|
limit 10
|
|
---- RESULTS
|
|
9999,false,999
|
|
9998,true,998
|
|
9997,false,997
|
|
9996,true,996
|
|
9995,false,995
|
|
9994,true,994
|
|
9993,false,993
|
|
9992,true,992
|
|
9991,false,991
|
|
9990,true,990
|
|
---- TYPES
|
|
INT, BOOLEAN, INT
|
|
====
|
|
---- QUERY
|
|
# The next sequence of queries is a regression test for IMPALA-4153
|
|
# verifying the retrieval of empty and NULL string columns
|
|
select count(*)
|
|
from nulltable
|
|
---- RESULTS
|
|
1
|
|
---- TYPES
|
|
BIGINT
|
|
====
|
|
---- QUERY
|
|
select count(*)
|
|
from nulltable where b = ''
|
|
---- RESULTS
|
|
1
|
|
---- TYPES
|
|
BIGINT
|
|
====
|
|
---- QUERY
|
|
select a,b
|
|
from nulltable where b = ''
|
|
---- RESULTS
|
|
'a',''
|
|
---- TYPES
|
|
STRING, STRING
|
|
====
|
|
---- QUERY
|
|
# The following 3 tests are regression tests for IMPALA-6187. Make sure the conjuncts are
|
|
# evaluated when there are no materialized slots or only partition columns are accessed.
|
|
select count(*) from alltypes where rand() * 10 >= 0.0;
|
|
---- RESULTS
|
|
7300
|
|
---- TYPES
|
|
BIGINT
|
|
====
|
|
---- QUERY
|
|
select count(*) from alltypes where rand() * 10 < 0.0;
|
|
---- RESULTS
|
|
0
|
|
---- TYPES
|
|
BIGINT
|
|
====
|
|
---- QUERY
|
|
# 'year' and 'month' are partition columns.
|
|
select count(*) from alltypes where rand() - year > month;
|
|
---- RESULTS
|
|
0
|
|
---- TYPES
|
|
BIGINT
|
|
====
|