[CALCITE-5023] If the inputs to a join have duplicate filed names, JDBC adapter should not generate “SELECT *”

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.29.0
• Component/s: jdbc-adapter
• Labels: pull-request-available

Environment

jdk8

• Assignee: yanjing.wang
• Reporter: yanjing.wang
• Created: 25/Feb/22 06:26
• Updated: 28/Feb/22 19:50

Description

@Test void testJoinInputsProjectingSameFieldName() {
  final String query = "SELECT \"t\".\"id\", \"t0\".\"id\" AS \"id0\"\n"
      + "FROM (VALUES (NULL)) AS \"t\" (\"id\"),\n"
      + "(VALUES (NULL)) AS \"t0\" (\"id\")";
  final String expected = query;
  sql(query).ok(expected);
} 

The Join left and right have same field name id, we should generate an expected statement which equals to query, but Calcite generates SELECT * rather SELECT expression list.

 

In many dialects such as MYSQL, SPARK SQL, SELECT * which contains same field name can't be used in CTAS, for example

CREATE TABLE SELECT * FROM (VALUES (NULL)) AS \"t\" (\"id\"),
(VALUES (NULL)) AS \"t0\" (\"id\")

This will cause exception: Found duplicate column(s) in the table definition.

 

Issue Links

• links to: GitHub Pull Request #2731