[
https://issues.apache.org/jira/browse/CALCITE-3503?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16978249#comment-16978249
]
Danny Chen commented on CALCITE-3503:
-------------------------------------
Thanks [~xndai] for the update PR ! But i still have some concerns ~
The isValid check is always invoked from a RelOptRuleCall cycle(except the
fresh new root node), so i think fetch a RMQ from the new registered node is
safe.
We do not need to get a cluster from the RelSubSet.best, we can make the method
{code:java}
isValid(RelMetadataQueqy rmq, Litmus litmus) // to have a explicit RMQ as
argument
{code}
And there is always a RMQ from the new registered relational node.
> NPE at VolcanoPlanner#isValid when DEBUG is enabled
> ---------------------------------------------------
>
> Key: CALCITE-3503
> URL: https://issues.apache.org/jira/browse/CALCITE-3503
> Project: Calcite
> Issue Type: Bug
> Components: core
> Affects Versions: 1.21.0
> Reporter: Stamatis Zampetakis
> Priority: Blocker
> Labels: pull-request-available
> Fix For: 1.22.0
>
> Time Spent: 3h
> Remaining Estimate: 0h
>
> The stacktrace is the following:
> {noformat}
> java.lang.NullPointerException
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.isValid(VolcanoPlanner.java:880)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.ensureRegistered(VolcanoPlanner.java:870)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.ensureRegistered(VolcanoPlanner.java:92)
> at
> org.apache.calcite.rel.AbstractRelNode.onRegister(AbstractRelNode.java:321)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.registerImpl(VolcanoPlanner.java:1701)
> at
> org.apache.calcite.plan.volcano.VolcanoPlanner.setRoot(VolcanoPlanner.java:298)
> at
> org.apache.calcite.tools.Programs.lambda$standard$3(Programs.java:269)
> {noformat}
> According to [~volodymyr] this is a regression caused by CALCITE-3487.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
