pick from master #33979
commit id 65fb7d43b7e838c48502d4e8a69e2541dc73aa88
This PR:
1. add a new Command type: UnsupportedCommand to handle the statement only support parse but could not execute.
2. support syntax about set and use
3. add keyword VAULT to follow legacy planner
TODO
1. support all statment syntax in Nereids
In MySQL, it's common to use a simplified syntax like `SELECT constant FROM dual`
which is equivalent to just `SELECT constant`.
This syntax is often used by BI tools when utilizing MySQL connectors to verify connection validity.
To enhance compatibility and ensure seamless integration with such tools,
we have now implemented this feature in Doris.
### Key Changes:
- Doris now interprets `SELECT constant FROM dual` as `SELECT constant`, aligning with MySQL's behavior.
- This update ensures that BI tools can use standard MySQL connectors without modifications or errors when connecting to Doris.
In SQL syntax, `col != ''` equals `col.length() > 0`.
It means that this column must exist in ES doc fields and its content is not empty.
In this PR, we make a special translation for this binary predicate to keep the behavior of both consistent.
---------
Co-authored-by: Luennng <luennng@gmail.com>
Record increase refersh version more accurately.
The refreshVersion in the memo will increase when mv rewrite successfully.
In query rewrite, if refresh version is different from the current struct info map in group,
will refresh the group struct info or not.
Expand bucket shuffle downgrade condition, which originally requiring a single partition after pruning, basic table and bucket number < para number. Currently, we expect this option can be used for disabling bucket shuffle more efficiently, without above restrictions.
Co-authored-by: zhongjian.xzj <zhongjian.xzj@zhongjianxzjdeMacBook-Pro.local>
we save bi-map in cte consumer to get the maping between producer and consumer.
the consumer's output is decided by the map in it.
so, cte consumer should be output prunable, and should remove useless entry from map when do column pruning
AUTO PARTITION grammar has changed since #31585, but the output
of SHOW CREATE TABLE was left out to change, so the result is not
able to be recognized by the FE parser.
Fix exception when create materialized view with cte, after this fix, can create materialized view with following
```
CREATE MATERIALIZED VIEW mv_with_cte
BUILD IMMEDIATE REFRESH AUTO ON MANUAL
DISTRIBUTED BY RANDOM BUCKETS 2
PROPERTIES ('replication_num' = '1')
AS
with `test_with` AS (
select l_partkey, l_suppkey
from lineitem
union
select
ps_partkey, ps_suppkey
from
partsupp)
select * from test_with;
```
this is brought from https://github.com/apache/doris/pull/28144
mysql> grant SELECT_PRIV on `_internal_schema`.* to 'test'@'%'; ERROR 1102 (42000): errCode = 2, detailMessage = Incorrect database name '_internal_schema'
* Adapt paimon 0.6.0 (#33943)
Version 2.0.0 of the shade package eliminates potential jar conflicts, resolves dependency component issues, and significantly reduces package size.
Utilize the directly-dependent guava library instead of relying on transitively included libraries.
* [chore](dependencies)Upgrade paimon to 0.7.0 (#33987)
---------
Co-authored-by: Calvin Kirs <kirs@apache.org>
We should tell the user the correct error message when some thing wrong. But error message is in a mess. I will make it clear.
This is the first step: unify the error code usage in FE.
