[
https://issues.apache.org/jira/browse/CALCITE-2598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16641224#comment-16641224
]
Julian Hyde commented on CALCITE-2598:
--------------------------------------
Here's the last line of the stack:
{noformat}
Suppressed: org.apache.calcite.test.fuzzer.RexProgramFuzzyTest$1: seed
-5040647096973595653
{noformat}
I wasn't able to reproduce the issue using that seed. It was generated after a
long run with startSeed 0 and over 200k iterations.
> Failure in RexProgramFuzzyTest
> ------------------------------
>
> Key: CALCITE-2598
> URL: https://issues.apache.org/jira/browse/CALCITE-2598
> Project: Calcite
> Issue Type: Task
> Reporter: Julian Hyde
> Assignee: Julian Hyde
> Priority: Major
>
> I got the following failure in RexProgramFuzzyTest. Is it a duplicate issue?
> Failure #1
> {noformat}
> 2018-09-28 03:52:59,737 [pool-1-thread-1] INFO - Using seed 44 for rex
> fuzzing
> 2018-09-28 03:52:59,748 [pool-1-thread-1] INFO - Rex fuzzing results: number
> of cases tested=18, failed cases=0, duplicate failures=0, fuzz rate=1636 per
> second
> 2018-09-28 03:52:59,751 [pool-1-thread-1] INFO - Rex fuzzing results: number
> of cases tested=2, failed cases=1, duplicate failures=0, fuzz rate=666 per
> second
> 2018-09-28 03:52:59,754 [pool-1-thread-1] INFO - Randomized test identified
> a potential defect. Feel free to fix that issue
> java.lang.AssertionError: COALESCE(CASE(IS NOT FALSE(true),
> -(COALESCE(CAST(-1):INTEGER, null, ?0.notNullInt1), *(CAST(100500):INTEGER,
> ?0.notNullInt0)), -(+(-1990568273))), +(-(COALESCE(null, ?0.notNullInt0,
> ?0.notNullInt0))))
> coalesce(case_(isNotFalse(trueLiteral),
> sub(coalesce(rexBuilder.makeCall(SqlStdOperatorTable.CAST, literal(-1)),
> nullInt, vIntNotNull(1)), rexBuilder.makeCall(SqlStdOperatorTable.*,
> rexBuilder.makeCall(SqlStdOperatorTable.CAST, literal(100500)),
> vIntNotNull(0))), unaryMinus(unaryPlus(literal(-1990568273)))),
> unaryPlus(unaryMinus(coalesce(nullInt, vIntNotNull(0), vIntNotNull(0))))) had
> non-nullable type INTEGER, and it was optimized to CASE(IS NOT FALSE(true),
> -(COALESCE(CAST(-1):INTEGER, null, ?0.notNullInt1), *(CAST(100500):INTEGER,
> ?0.notNullInt0)), -(+(-1990568273)))
> case_(isNotFalse(trueLiteral),
> sub(coalesce(rexBuilder.makeCall(SqlStdOperatorTable.CAST, literal(-1)),
> nullInt, vIntNotNull(1)), rexBuilder.makeCall(SqlStdOperatorTable.*,
> rexBuilder.makeCall(SqlStdOperatorTable.CAST, literal(100500)),
> vIntNotNull(0))), unaryMinus(unaryPlus(literal(-1990568273)))) that has
> nullable type INTEGER, unknownAsFalse {noformat}
> I suggest that we append new failures to this case. If they are new issues,
> log new cases, but there is also a good chance that they are variants of, or
> related to, other open cases.
> In this case, it could be one of the two COALESCE issues already open.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
