[
https://issues.apache.org/jira/browse/CALCITE-4633?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17357580#comment-17357580
]
Sean Broeder commented on CALCITE-4633:
---------------------------------------
I think this query demonstrates an opportunity for improvement:
select count ("customer_id") from
(select "sales1"."customer_id", "sales1"."store_id", "sales1"."unit_sales",
"store1"."store_name"
from "sales_fact_1997" as "sales1"
inner join "store" as "store1" on "sales1"."store_id" = "store1"."store_id"
union
select "sales2"."customer_id", "sales2"."store_id", "sales2"."unit_sales",
"store2"."store_name"
from "sales_fact_1998" as "sales2"
inner join "store" as "store2" on "sales2"."store_id" = "store2"."store_id")
The inputs for the union are joins and below the joins we do a complete table
scan without trimming. With the change I am proposing there would be a project
between the table scan and the join, which should improve performance.
Please let me know your thoughts and if you agree it is worth review I will
push a PR for evaluation.
> Enable RelFieldTrimmer to trim fields on a distinct UNION
> ---------------------------------------------------------
>
> Key: CALCITE-4633
> URL: https://issues.apache.org/jira/browse/CALCITE-4633
> Project: Calcite
> Issue Type: Improvement
> Components: core
> Affects Versions: 1.26.0
> Reporter: Sean Broeder
> Assignee: Sean Broeder
> Priority: Major
> Fix For: 1.28.0
>
>
> Should be able to push a project to trim fields on top of distinct union
> without changing results.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
