[ 
https://issues.apache.org/jira/browse/CALCITE-4160?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17172777#comment-17172777
 ] 

Jiatao Tao edited comment on CALCITE-4160 at 8/7/20, 2:51 AM:
--------------------------------------------------------------

Hi [~zabetak], [~julianhyde]

In ByteDance, we use Calcite as the SQL layer(parser/validate/optimizer) to 
build a Federated Query System like google F1. So it needs to adapt to the 
various engines like Spark/Presto/ClickHouse etc.

Take Spark, for example, we need to use RelNode to construct DataFrame, and 
Spark allows sort in sub-query. Take other engines, we convert RelNode back to 
SQL, whether they can handle sort in sub-query is by themself, but the SQL 
should have this info.

 

In fact, Presto/ClickHouse support sort in sub-query too, so they want us to 
keep this in SQL.
  


was (Author: aron.tao):
Hi [~zabetak], [~julianhyde]

In ByteDance, we use Calcite as the SQL layer(parser/validate/optimizer) to 
build a Federated Query System like google F1. So it needs to adapt to the 
various engines like Spark/Presto/ClickHouse.

Take Spark, for example, we need to use RelNode to construct DataFrame, and 
Spark allows sort in sub-query. Take other engines, we convert RelNode back to 
SQL, whether they can handle sort in sub-query is by themself, but the SQL 
should have this info.

 

In fact, Presto/ClickHouse support sort in sub-query too, so they want us to 
keep this in SQL.
  

> Add configuration to retain ORDER BY in sub-query
> -------------------------------------------------
>
>                 Key: CALCITE-4160
>                 URL: https://issues.apache.org/jira/browse/CALCITE-4160
>             Project: Calcite
>          Issue Type: Improvement
>          Components: core
>            Reporter: Jiatao Tao
>            Assignee: Jiatao Tao
>            Priority: Minor
>         Attachments: image-2020-08-07-10-48-59-599.png, 
> image-2020-08-07-10-50-22-351.png
>
>
> Pre discussion see in 
> https://issues.apache.org/jira/browse/CALCITE-2798
> https://issues.apache.org/jira/browse/CALCITE-3664
> IMO, we should at least provide a configuration about this. People use 
> Calcite most for its optimizer, and this info users may care, How it executed 
> is by the engine, but users should know this in the plan.
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to