RelBuilder.call deals with a scalar function, and I think what you need is a
table function. (In java both are just functions, but in SQL, or relational
algebra, they are different beasts.)
Calcite has a TableFunctionScan, which is a RelNode, takes a function, 0 or
more RelNode inputs, and 0 or more other arguments.
RelBuilder does not support TableFunctionScan yet. Can you please log a JIRA
case for that? (And implement it, if you are so inclined.)
You don’t need to wait for support to be added to RelBuilder. You can create a
TableFunctionScan manually, and send it to RelBuilder.push().
Julian
> On Nov 28, 2016, at 4:52 AM, Nikolay Vasilishin <nikolay_vasilis...@epam.com>
> wrote:
>
> Hi, devs.
> I'm trying to implement RexVisitor#visitSubquery method for Flink Table API.
> Can you help me with a problem?
>
>
> 1. I have to construct a RexNode while having a RexNode with left side
> of IN (name of column) and RelNode, which stands for table or subquery on the
> right side of IN operator. If I'm using RexSubQuery.in(rel, rex), I'm failing
> with exception:
>
>
>
> java.lang.ClassCastException: org.apache.calcite.rex.RexSubQuery cannot be
> cast to org.apache.calcite.rex.RexLocalRef
>
>
>
> at
> org.apache.calcite.rex.RexProgramBuilder.registerInput(RexProgramBuilder.java:291)
>
> at
> org.apache.calcite.rex.RexProgramBuilder.addCondition(RexProgramBuilder.java:265)
>
> at
> org.apache.calcite.rel.rules.FilterToCalcRule.onMatch(FilterToCalcRule.java:66)
>
> at
> org.apache.calcite.plan.volcano.VolcanoRuleCall.onMatch(VolcanoRuleCall.java:213)
>
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.findBestExp(VolcanoPlanner.java:819)
>
> at
> org.apache.calcite.tools.Programs$RuleSetProgram.run(Programs.java:334)
>
> at
> org.apache.flink.api.table.BatchTableEnvironment.optimize(BatchTableEnvironment.scala:252)
>
> at
> org.apache.flink.api.table.BatchTableEnvironment.translate(BatchTableEnvironment.scala:287)
>
> at
> org.apache.flink.api.scala.table.BatchTableEnvironment.toDataSet(BatchTableEnvironment.scala:139)
>
> at
> org.apache.flink.api.scala.table.TableConversions.toDataSet(TableConversions.scala:41)
>
> at
> org.apache.flink.api.scala.batch.table.InITCase.testInSubquery(InITCase.scala:72)
>
>
>
> RelBuilder#call doesn't take RelNode as parameter. Combining RelBuilder#call
> with output of RexSubQuery.in(rel, rex) as RexNode causes duplicated IN
> operator in logical plan:
>
>
>
> LogicalFilter(condition=[IN(IN($2, {
>
> LogicalProject(c=[$2])
>
> LogicalFilter(condition=[=($1, 6)])
>
> LogicalTableScan(table=[[_DataSetTable_0]])
>
> }))])
>
> LogicalTableScan(table=[[_DataSetTable_0]])
>
>
>
> Besides, these classes look very similar.
> So, the question is how to construct proper IN RexNode or how to cast from
> RexSubQuery to RexNode.
>
> Thanks in advance,
>
> Nikolay Vasilishin
> Junior Software Engineer
>
> Office: +7 812 611 10 94<tel:+7%20812%20611%2010%2094> x 61997<tel:61997>
> Cell: +7 921 403 78 55<tel:+7%20921%20403%2078%2055> Email:
> nikolay_vasilis...@epam.com<mailto:nikolay_vasilis...@epam.com>
> Saint-Petersburg, Russia (GMT+3) epam.com<http://www.epam.com>
>
> CONFIDENTIALITY CAUTION AND DISCLAIMER
> This message is intended only for the use of the individual(s) or entity(ies)
> to which it is addressed and contains information that is legally privileged
> and confidential. If you are not the intended recipient, or the person
> responsible for delivering the message to the intended recipient, you are
> hereby notified that any dissemination, distribution or copying of this
> communication is strictly prohibited. All unintended recipients are obliged
> to delete this message and destroy any printed copies.
>
