[
https://issues.apache.org/jira/browse/FLINK-6196?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15944401#comment-15944401
]
ASF GitHub Bot commented on FLINK-6196:
---------------------------------------
Github user KurtYoung commented on a diff in the pull request:
https://github.com/apache/flink/pull/3623#discussion_r108320419
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/schema/FlinkTableFunctionImpl.scala
---
@@ -27,52 +27,73 @@ import
org.apache.flink.api.common.typeinfo.{AtomicType, TypeInformation}
import org.apache.flink.api.common.typeutils.CompositeType
import org.apache.flink.table.api.TableException
import org.apache.flink.table.calcite.FlinkTypeFactory
+import org.apache.flink.table.functions.utils.UserDefinedFunctionUtils
+import org.apache.flink.table.functions.{TableFunction => FlinkUDTF}
/**
* This is heavily inspired by Calcite's
[[org.apache.calcite.schema.impl.TableFunctionImpl]].
* We need it in order to create a
[[org.apache.flink.table.functions.utils.TableSqlFunction]].
* The main difference is that we override the [[getRowType()]] and
[[getElementType()]].
+ *
+ * @param tableFunction The Table Function instance
+ * @param implicitResultType The implicit result type.
+ * @param evalMethod The eval() method of the [[tableFunction]]
*/
-class FlinkTableFunctionImpl[T](
- val typeInfo: TypeInformation[T],
- val fieldIndexes: Array[Int],
- val fieldNames: Array[String],
- val evalMethod: Method)
+class FlinkTableFunctionImpl[T](val tableFunction: FlinkUDTF[_],
--- End diff --
Can we just separate this implementation into 2 classes, one for which we
already know the type, and one for whose type has not been decided yet.
Why i suggesting this is that i noticed `tableFunction.getResultType` is
called multiple times when evaluating the plan, even we had already determine
the row type long time ago, i.e. when `apply` method from `TableFunction` has
been called.
> Support dynamic schema in Table Function
> ----------------------------------------
>
> Key: FLINK-6196
> URL: https://issues.apache.org/jira/browse/FLINK-6196
> Project: Flink
> Issue Type: Improvement
> Components: Table API & SQL
> Reporter: Zhuoluo Yang
> Assignee: Zhuoluo Yang
>
> In many of our use cases. We have to decide the schema of a UDTF at the run
> time. For example. udtf('c1, c2, c3') will generate three columns for a
> lateral view.
> Most systems such as calcite and hive support this feature. However, the
> current implementation of flink didn't implement the feature correctly.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
