modify the bind logical to make the order by has same behavior with mysql when sort child is aggregate.
when an order by Expr has aggregate function, all slots in this order by Expr should bind the LogicalAggregate non-AggFunction outputs first, then bind the LogicalAggregate Child
e.g.
select 2*abs(sum(c1)) as c1, c1,sum(c1)+c1 from t_order_by_bind_priority group by c1 order by sum(c1)+c1 asc;
in this sql, the two c1 in order by all bind to the c1 in t_order_by_bind_priority
with group by:
select max(1) from t1 group by c1; -> select 1 from (select c1 from t1 group by c1);
without group by:
select max(1) from t1; -> select max(1) from (select 1 from t1 limit 1) tmp;
This sql will failed because
2 in the group by will bind to 1 as col2 in BindExpression
ResolveOrdinalInOrderByAndGroupBy will replace 1 to MIN (LENGTH (cast(age as varchar)))
CheckAnalysis will throw an exception because group by can not contains aggregate function
select MIN (LENGTH (cast(age as varchar))), 1 AS col2
from test_bind_groupby_slots
group by 2
we should move ResolveOrdinalInOrderByAndGroupBy into BindExpression
(cherry picked from commit 3fab4496c3fefe95b4db01f300bf747080bfc3d8)
Fix failed in regression_test/suites/query_p0/group_concat/test_group_concat.groovy
select
group_concat( distinct b1, '?'), group_concat( distinct b3, '?')
from
table_group_concat
group by
b2
exception:
lowestCostPlans with physicalProperties(GATHER) doesn't exist in root group
The root cause is '?' is push down to slot by NormalizeAggregate, AggregateStrategies treat the slot as a distinct parameter and generate a invalid PhysicalHashAggregate, and then reject by ChildOutputPropertyDeriver.
I fix this bug by avoid push down literal to slot in NormalizeAggregate, and forbidden generate stream aggregate node when group by slots is empty
#32617 introduce a bug: rewrite may not working when plan's arity >= 3.
this pr fix it
(cherry picked from commit 8b070d1a9d43aa7d25225a79da81573c384ee825)
this pr can improve the performance of the nereids planner, in plan stage.
1. refactor expression rewriter to pattern match, so the lots of expression rewrite rules can criss-crossed apply in a big bottom-up iteration, and rewrite until the expression became stable. now we can process more cases because original there has no loop, and sometimes only process the top expression, like `SimplifyArithmeticRule`.
2. replace `Collection.stream()` to `ImmutableXxx.Builder` to avoid useless method call
3. loop unrolling some codes, like `Expression.<init>`, `PlanTreeRewriteBottomUpJob.pushChildrenJobs`
4. use type/arity specified-code, like `OneRangePartitionEvaluator.toNereidsLiterals()`, `PartitionRangeExpander.tryExpandRange()`, `PartitionRangeExpander.enumerableCount()`
5. refactor `ExtractCommonFactorRule`, now we can extract more cases, and I fix the deed loop when use `ExtractCommonFactorRule` and `SimplifyRange` in one iterative, because `SimplifyRange` generate right deep tree, but `ExtractCommonFactorRule` generate left deep tree
6. refactor `FoldConstantRuleOnFE`, support visitor/pattern match mode, in ExpressionNormalization, pattern match can criss-crossed apply with other rules; in PartitionPruner, visitor can evaluate expression faster
7. lazy compute and cache some operation
8. use int field to compare date
9. use BitSet to find disableNereidsRules
10. two level loop usually faster then build Multimap when bind slot in Scope, so I revert the code
11. `PlanTreeRewriteBottomUpJob` don't need to clearStatePhase any more
### test case
100 threads parallel continuous send this sql which query an empty table, test in my mac machine(m2 chip, 8 core), enable sql cache
```sql
select count(1),date_format(time_col,'%Y%m%d'),varchar_col1
from tbl
where partition_date>'2024-02-15' and (varchar_col2 ='73130' or varchar_col3='73130') and time_col>'2024-03-04'
and time_col<'2024-03-05'
group by date_format(time_col,'%Y%m%d'),varchar_col1
order by date_format(time_col,'%Y%m%d') desc, varchar_col1 desc,count(1) asc
limit 1000
```
before this pr: 3100 peak QPS, about 2700 avg QPS
after this pr: 4800 peak QPS, about 4400 avg QPS
(cherry picked from commit 7338683fdbdf77711f2ce61e580c19f4ea100723)
Previously, strings_pool was allocated within each tree node. However, due to the Arena's alignment of allocated chunks to at least 4K, this allocation size was excessively large for a single tree node. Consequently, when there are numerous nodes within the SubcolumnTree, a significant portion of memory was wasted. Moving strings_pool to the tree itself optimizes memory usage and reduces wastage, improving overall efficiency.
* The default delete bitmap cache is set to 100MB, which can be insufficient and cause performance issues when the amount of user data is large. To mitigate the problem of an inadequate cache, we will take the larger of 5% of the total memory and 100MB as the delete bitmap cache size.
