Hi community,

I'm trying to convert a plan to a sql of SPARK dialect, but sometimes the
fields will be converted to star, see the following example

*create table users(id int);*
*create table depts(id int);*

*select a.id <http://a.id>, b.id <http://b.id> as id0 from users1 a, depts1
b*

converting to plan results
LogicalProject(id=[$0], id0=[$1])
  LogicalJoin(condition=[true], joinType=[inner])
    HiveTableScan(table=[[default, users1]])
    HiveTableScan(table=[[default, depts1]])

converting the plan to SPARK sql results


*SELECT *FROM `default`.`users1`CROSS JOIN `default`.`depts1`*

because PROJECT and JOIN row type are identical, the *PROJECT* has been
converted to *SELECT ** .

the problem is the SPARK sql can't be used to create table, such as


*CREATE TABLE tmp AS SELECT *FROM `default`.`users1`CROSS JOIN
`default`.`depts1`*

this ctas will throw  '*Found duplicate column(s) in the table definition'*

I wonder if we can add a config to indicate the *PROJECT *couldn't be
converted to star.

how do you think?

Reply via email to