[
https://issues.apache.org/jira/browse/FLINK-27247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17524251#comment-17524251
]
Marios Trivyzas commented on FLINK-27247:
-----------------------------------------
We can fix the issue with the quick fix in your PR.
We have to open an issue, so that these kind of explicit casting is introduced
with a Rule during the analysis/planning phase, and not in the code generation.
This way, it will be visible what kind of implicit cast is introduced, by
looking at the plan, and not just generate code for it under the hood.
> ScalarOperatorGens.numericCasting is not compatible with legacy behavior
> ------------------------------------------------------------------------
>
> Key: FLINK-27247
> URL: https://issues.apache.org/jira/browse/FLINK-27247
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Reporter: xuyang
> Priority: Minor
> Labels: pull-request-available
>
> Add the following test cases in ScalarFunctionsTest:
> {code:java}
> // code placeholder
> @Test
> def test(): Unit ={
> testSqlApi("rand(1) + 1","")
> } {code}
> it will throw the following exception:
> {code:java}
> // code placeholder
> org.apache.flink.table.planner.codegen.CodeGenException: Unsupported casting
> from DOUBLE to DOUBLE NOT NULL.
> at
> org.apache.flink.table.planner.codegen.calls.ScalarOperatorGens$.numericCasting(ScalarOperatorGens.scala:1734)
> at
> org.apache.flink.table.planner.codegen.calls.ScalarOperatorGens$.generateBinaryArithmeticOperator(ScalarOperatorGens.scala:85)
> at
> org.apache.flink.table.planner.codegen.ExprCodeGenerator.generateCallExpression(ExprCodeGenerator.scala:507)
> at
> org.apache.flink.table.planner.codegen.ExprCodeGenerator.visitCall(ExprCodeGenerator.scala:481)
> at
> org.apache.flink.table.planner.codegen.ExprCodeGenerator.visitCall(ExprCodeGenerator.scala:57)
> at org.apache.calcite.rex.RexCall.accept(RexCall.java:174)
> at
> org.apache.flink.table.planner.codegen.ExprCodeGenerator.$anonfun$visitCall$1(ExprCodeGenerator.scala:478)
> at
> scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:233)
> at
> scala.collection.mutable.ResizableArray.foreach(ResizableArray.scala:58)
> at
> scala.collection.mutable.ResizableArray.foreach$(ResizableArray.scala:51)
> at scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:47)
> at scala.collection.TraversableLike.map(TraversableLike.scala:233)
> at scala.collection.TraversableLike.map$(TraversableLike.scala:226)
> at scala.collection.AbstractTraversable.map(Traversable.scala:104)
> at
> org.apache.flink.table.planner.codegen.ExprCodeGenerator.visitCall(ExprCodeGenerator.scala:469)
> ... {code}
> This is because in ScalarOperatorGens#numericCasting, FLINK-24779 lost the
> logic that in some cases there is no need to casting the left and right type.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
