Problem:
When create view with join in table partitions, an error would rise like "Unknown column"
Example:
CREATE VIEW my_view AS SELECT t1.* FROM t1 PARTITION(p1) JOIN t2 PARTITION(p2) ON t1.k1 = t2.k1;
select * from my_view ==> errCode = 2, detailMessage = Unknown column 'k1' in 't2'
Reason:
When create view, we do tosql first in order to persistent view sql. And when doing tosql of table reference, partition key
word was removed to keep neat of sql string. But here when we remove partition keyword it would regarded as an alias.
So "PARTITION" keyword can not be removed.
Solved:
Add “PARTITION” keyword back to tosql string.
join commute rule will swap the left and right child. This cause the change of logical properties. So we need recompute the logical properties in plan post process to get the correct result
Problem:
When create view with projection group_concat(xxx, xxx order by orderkey). It will failed during second parse of inline view
For example:
it works when doing
"SELECT id, group_concat(`name`, "," ORDER BY id) AS test_group_column FROM test GROUP BY id"
but when create view it does not work
"create view test_view as SELECT id, group_concat(`name`, "," ORDER BY id) AS test_group_column FROM test GROUP BY id"
Reason:
when creating view, we will doing parse again of view.toSql() to check whether it has some syntax error. And when doing toSql() to group_concat with order by, it add seperate ', ' between second parameter and order by. So when parsing again, it
would failed because it is different semantic with original statement.
group_concat(`name`, "," ORDER BY id) ==> group_concat(`name`, "," , ORDER BY id)
Solved:
Change toSql of group_concat and add order by statement analyze() of group_concat in Planner cause it would work if we get order by from view statement and do not analyze and binding slot reference to it
Cache the iceberg table. When accessing the same table, the metadata will only be loaded once.
Cache the snapshot of the table to optimize the performance of the iceberg table function.
Add cache support for iceberg's manifest file content
a simple test from 2.0s to 0.8s
before
mysql> refresh table tb3;
Query OK, 0 rows affected (0.03 sec)
mysql> select * from tb3;
+------+------+------+
| id | par | data |
+------+------+------+
| 1 | a | a |
| 2 | a | b |
| 3 | a | c |
....
| 68 | a | a |
| 69 | a | b |
| 70 | a | c |
+------+------+------+
70 rows in set (2.10 sec)
mysql> select * from tb3;
+------+------+------+
| id | par | data |
+------+------+------+
| 1 | a | a |
| 2 | a | b |
| 3 | a | c |
...
| 68 | a | a |
| 69 | a | b |
| 70 | a | c |
+------+------+------+
70 rows in set (2.00 sec)
after
mysql> refresh table tb3;
Query OK, 0 rows affected (0.03 sec)
mysql> select * from tb3;
+------+------+------+
| id | par | data |
+------+------+------+
| 1 | a | a |
| 2 | a | b |
...
| 68 | a | a |
| 69 | a | b |
| 70 | a | c |
+------+------+------+
70 rows in set (2.05 sec)
mysql> select * from tb3;
+------+------+------+
| id | par | data |
+------+------+------+
| 1 | a | a |
| 2 | a | b |
| 3 | a | c |
...
| 68 | a | a |
| 69 | a | b |
| 70 | a | c |
+------+------+------+
70 rows in set (0.80 sec)
When adding be, it is required to have only one colon, otherwise an error will be reported. However, ipv6 has many colons
```
String[] pair = hostPort.split(":");
if (pair.length != 2) {
throw new AnalysisException("Invalid host port: " + hostPort);
}
```
Optimization "select count(*) from table" stmtement , push down "count" type to BE.
support file type : parquet ，orc in hive .
1. 4kfiles , 60kwline num
before: 1 min 37.70 sec
after: 50.18 sec
2. 50files , 60kwline num
before: 1.12 sec
after: 0.82 sec
Enhance broadcast join cost calculation, by considering both the build side effort from building bigger hash table, and more probe side effort from bigger cost of ProbeWhenBuildSideOutput and ProbeWhenSearchHashTable, if parallel_fragment_exec_instance_num is more than 1.
Current solution gives a penalty factor on rightRowCount, and the factor is the total instance number to the power of 2.
Penalty on outputRows is not taken currently and will be refined in next generation cost model.
Also brings some update for shape checking:
update original control variable in shape file parallel_fragment_exec_instance_num to parallel_pipeline_task_num, if pipeline is enabled.
fix a be_number variable inactive issue.
consider sql:
```
SELECT *
FROM sub_query_correlated_subquery1 t1
WHERE coalesce(bitand(
cast(
(SELECT sum(k1)
FROM sub_query_correlated_subquery3 ) AS int),
cast(t1.k1 AS int)),
coalesce(t1.k1, t1.k2)) is NULL
ORDER BY t1.k1, t1.k2;
```
is Null conjunct is lost in SubqueryToApply rule. This pr fix it
After auto retry merged, it's hard to determine the execute times of doExecute method in compile time, and if the expected execute times in the expectation block is missed, unexpected invocation exception would be thrown, so just remove the expected execute times
select c_name from customer union select c_name from customer
this sql used agg node to get distinct row of c_name,
so it's no need to wait for inserted all data to hash map,
could output the data which it's inserted into hash map successed.
1. not use recursive parse to avoid stack overflow
2. To create a balanced tree instead of left deep tree
TODO: add expr_depth_limit to Nereids' parser
This file system cache key should contains `scheme://authority`, eg: `hdfs//nameservices1`.
Or it will encounter error:
```
Wrong FS: hdfs//abc/xxxx, expected: hdfs://def
```
