[
https://issues.apache.org/jira/browse/CALCITE-1222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15266615#comment-15266615
]
Kevin Risden commented on CALCITE-1222:
---------------------------------------
I tested some more queries and no others stood out. I think this change is good.
> DatabaseMetaData.getColumnLabel returns null when query has ORDER BY
> --------------------------------------------------------------------
>
> Key: CALCITE-1222
> URL: https://issues.apache.org/jira/browse/CALCITE-1222
> Project: Calcite
> Issue Type: Bug
> Reporter: Kevin Risden
> Assignee: Julian Hyde
> Fix For: 1.8.0
>
>
> Background from the thread here:
> http://mail-archives.apache.org/mod_mbox/calcite-dev/201604.mbox/%3CCAJU9nmgxyR%2BCXPXXjYkSKRL4%3DUvbN4nD1JMCtiFzAofRrT1Pug%40mail.gmail.com%3E
> [~risdenk] wrote:
> {quote}
> I've been integrating Calcite/Avatica into Solr and ran into a case where
> adding an order by or a limit changes the output of
> DatabaseMetaData.getColumnName and getColumnLabel.
> An example is below:
> Behavior 1 (which I think is correct from reading the JDBC DatabaseMetaData
> spec):
> - select e."name" as full_name from "hr"."emps" as e;
> - getColumnName(1) -> name
> - getColumnLabel(1) -> FULL_NAME
> Behavior 2:
> - select e."name" as full_name from "hr"."emps" as e limit 10;
> - getColumnName(1) -> FULL_NAME
> - getColumnLabel(1) -> FULL_NAME
> I was able to reproduce the behavior with the JdbcExample.java file in the
> master branch of Calcite since I wasn't sure if it was my adapter
> originally.
> I started looking at the SqlParserTest and added this test case which
> doesn't error out.
> /**
> * "LIMIT ... OFFSET ..." is the postgres equivalent of SQL:2008
> * "OFFSET ... FETCH". It all maps down to a parse tree that looks like
> * SQL:2008.
> */
> @Test public void testColumnAliasWithAsWithLimit() {
> check(
> "select a as abc from foo order by b, c limit 2 offset 1",
> "SELECT `A` AS `ABC`\n"
> + "FROM `FOO`\n"
> + "ORDER BY `B`, `C`\n"
> + "OFFSET 1 ROWS\n"
> + "FETCH NEXT 2 ROWS ONLY");
> check(
> "select a as abc from foo order by b, c limit 2",
> "SELECT `A` AS `ABC`\n"
> + "FROM `FOO`\n"
> + "ORDER BY `B`, `C`\n"
> + "FETCH NEXT 2 ROWS ONLY");
> check(
> "select a as abc from foo order by b, c offset 1",
> "SELECT `A` AS `ABC`\n"
> + "FROM `FOO`\n"
> + "ORDER BY `B`, `C`\n"
> + "OFFSET 1 ROWS");
> I'm not sure where to look next to try to find this to submit a patch. Let
> me know if you need more information.
> I think this potentially causes some problems down the line when the
> original name or alias is used for ordering and having, but I haven't been
> able to verify this for sure.
> {quote}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
