[
https://issues.apache.org/jira/browse/CALCITE-873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14730049#comment-14730049
]
Julian Hyde commented on CALCITE-873:
-------------------------------------
I have a hypothesis that the following rule is sufficient: "you can safely
remove from a sort key any columns known to be constant".
> Prevent sort when ORDER BY not necessary due to equality constraints
> --------------------------------------------------------------------
>
> Key: CALCITE-873
> URL: https://issues.apache.org/jira/browse/CALCITE-873
> Project: Calcite
> Issue Type: Improvement
> Reporter: James Taylor
> Assignee: Julian Hyde
>
> We're working on an optimization in Phoenix to optimize away an ORDER BY when
> it is known based on equality expressions in the WHERE clause that it is not
> necessary (PHOENIX-2194). It'd be great if Calcite could do that as well.
> Here's a example, given the following schema:
> {code}
> CREATE TABLE T (
> K1 VARCHAR,
> K2 VARCHAR,
> K3 VARCHAR,
> CONSTRAINT pk PRIMARY KEY (K1, K2, K3));
> {code}
> In the following queries, no sort is necessary:
> {code}
> SELECT * FROM T WHERE K1='A' ORDER BY K2,K3;
> SELECT * FROM T WHERE K2='B' ORDER BY K1,K3;
> SELECT * FROM T WHERE K1='A' AND K2='B' ORDER BY K3;
> {code}
> There are also some edge cases where a function may be known to select a
> *prefix* of the column value where it's still ok to not sort:
> {code}
> SELECT * FROM T WHERE K1='A' AND SUBSTR(K2,1,3)='ABC' ORDER BY K2;
> {code}
> But if another column is included in the ORDER BY after the prefixing, a sort
> would still be necessary:
> {code}
> SELECT * FROM T WHERE K1='A' AND SUBSTR(K2,1,3)='ABC' ORDER BY K2,K3;
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
