[jira] [Comment Edited] (CALCITE-5725) Support Doris Dialect

["xingyuan cheng (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CALCITE-5725?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17726560#comment-17726560 ] 

xingyuan cheng edited comment on CALCITE-5725 at 5/26/23 9:38 AM:
------------------------------------------------------------------

[~libenchao] 

Hello, first of all, thank you for your reply. The reason for this change is mainly based on two reasons: 1. Because we internally expect a tool for dialect conversion based on calcite, because there are some historical sql inside, like spark, hive expects to query When the speed is slow, switch to an MPP database such as doris for querying, and the Mysql dialect supported by doris itself has made some custom expansions inside. For example, the grammar extensions supported by DorisDB include DISTRIBUTE BY, CLUSTER BY, SORT BY , WINDOW, etc. These are not supported in the native Mysql syntax supported by calcite itself. In other words, even if the doris-related syntax extension is supported on the basis of the original Mysql dialect, the conversion of the Mysql dialect itself will be destroyed.

2. In the process of using Doris internally, since the data volume is maintained at about 6PB per day, the use of distributed tables is basically considered, so it is expected to support the doris dialect without destroying the complete semantics of mysql in calcite, and Attempt to complete various conversions from doris to other dialects, and from other dialects to doris.


was (Author: complone):
[~libenchao] 您好，首先感谢您的回复。之所以这么改，主要基于两个原因：

1.因为我们内部期待一个基于calcite的方言转换的工具，因为里面有一些历史sql，像spark，hive希望查询速度慢的时候，改用一个MPP数据库如doris用于查询，doris本身支持的Mysql方言在里面做了一些自定义扩展。比如DorisDB支持的语法扩展包括DISTRIBUTE BY、CLUSTER BY、SORT BY、WINDOW等，这些在calcite本身支持的原生Mysql语法中是不支持的。也就是说，即使在原有Mysql方言的基础上支持doris相关的语法扩展，也会破坏Mysql方言本身的转换。

2、在内部使用Doris的过程中，由于数据量维持在每天6PB左右，基本考虑使用分布式表，所以希望在不破坏calcite中mysql完整语义的情况下支持doris方言，尝试完成从 doris 到其他方言以及从其他方言到 doris 的各种转换。

> Support Doris Dialect
> ---------------------
>
>                 Key: CALCITE-5725
>                 URL: https://issues.apache.org/jira/browse/CALCITE-5725
>             Project: Calcite
>          Issue Type: New Feature
>          Components: core
>            Reporter: xingyuan cheng
>            Priority: Major
>
> Doris is one of the popular OLAP databases. The current version of calcite does not support the dialect of Doris as input syntax, and it is also unable to perform dialect conversion based on Doris. For example, doris is converted to hive, doris is converted to flink, and spark is converted to doris



--
This message was sent by Atlassian Jira
(v8.20.10#820010)