CI Red: Fix union in view table test #15300
Conversation
| 07)----SortExec: expr=[y@0 ASC NULLS LAST], preserve_partitioning=[false] | ||
| 08)------DataSourceExec: partitions=1, partition_sizes=[1] | ||
|
|
||
| # optimize_subquery_sort in create_relation removes Sort so the result is not sorted. |
There was a problem hiding this comment.
I have no idea why we have optimize plan rule optimize_subquery_sort in create_relation, I think we should move such rule to optimizer 🤔
There was a problem hiding this comment.
Before #15201 - This plan doesn't meet the optimize rule condition because we didn't inline view table at this point, so the Sort is not removed.
Now - The view table inlined so Sort is removed by this function.
There was a problem hiding this comment.
I think removing Sort makes sense, but we need to move this rule to optimizer.
There was a problem hiding this comment.
I agree using ORDER BY in a view is meaningless.
jayzhan211
changed the title
| 07)----SortExec: expr=[y@0 ASC NULLS LAST], preserve_partitioning=[false] | ||
| 08)------DataSourceExec: partitions=1, partition_sizes=[1] | ||
|
|
||
| # optimize_subquery_sort in create_relation removes Sort so the result is not sorted. |
There was a problem hiding this comment.
I agree using ORDER BY in a view is meaningless.
|
|
||
| # optimize_subquery_sort in create_relation removes Sort so the result is not sorted. | ||
| query I | ||
| SELECT * FROM v1; |
There was a problem hiding this comment.
In this Select query, Postgre does not drop the Sort operation, FYI @jayzhan211 @jonahgao
There was a problem hiding this comment.
I think either is fine. A view is a type of relation, and a relation is unordered.
In the SQL Server doc, it states
The ORDER BY clause does not guarantee ordered results when the view is queried, unless ORDER BY is also specified in the query itself.
Postgres also does not drop the Sort operation in subqueries but optimize_subquery_sort does.
psql=> explain select * from (select from u1 order by x);
QUERY PLAN
----------------------------------------------------------------------------
Subquery Scan on unnamed_subquery (cost=158.51..186.76 rows=2260 width=0)
-> Sort (cost=158.51..164.16 rows=2260 width=4)
Sort Key: u1.x
-> Seq Scan on u1 (cost=0.00..32.60 rows=2260 width=4)
(4 rows)
3 participants
Which issue does this PR close?
This CI failure is caused by the conflict with #15135, so is not detected
Rationale for this change
What changes are included in this PR?
Are these changes tested?
Are there any user-facing changes?