[ 
https://issues.apache.org/jira/browse/CALCITE-3052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16836996#comment-16836996
 ] 

Anton Haidai commented on CALCITE-3052:
---------------------------------------

Other than that, looks like I found two other potential bugs while trying to 
reproduce the issue in "MaterializationTest":
h2. Issue 1: target out of range
{code}
  @Test public void testPremutationError() {
    CalciteAssert.that()
        .withMaterializations(
            HR_FKUK_MODEL,
            "m0",
            "select min(\"salary\"), count(*), max(\"salary\"), 
sum(\"salary\"), \"empid\" from \"emps\" group by \"empid\"",
            "m1",
            "select min(\"salary\"), count(*), max(\"salary\"), 
sum(\"salary\"), \"deptno\", \"empid\" from \"emps\" group by \"empid\", 
\"deptno\""
        )
        .query(
            "select count(*), \"empid\" from \"emps\" group by \"empid\"")
        .enableMaterializations(true)
        .explainContains("EnumerableTableScan(table=[[hr, m0]])")
        .sameResultWithMaterializationsDisabled();
  }
{code}
Error (looks like the mapping is expected to be bijection but it is not): 
{code}
Caused by: java.lang.IllegalArgumentException: target out of range
        at org.apache.calcite.util.Permutation.<init>(Permutation.java:69)
        at org.apache.calcite.util.mapping.Mappings.bijection(Mappings.java:394)
        at 
org.apache.calcite.rel.mutable.MutableRels.createProject(MutableRels.java:142)
        at 
org.apache.calcite.plan.SubstitutionVisitor.unifyAggregates(SubstitutionVisitor.java:1269)
        at 
org.apache.calcite.plan.SubstitutionVisitor$AggregateOnProjectToAggregateUnifyRule.apply(SubstitutionVisitor.java:1345)
        at 
org.apache.calcite.plan.SubstitutionVisitor.go(SubstitutionVisitor.java:531)
        at 
org.apache.calcite.plan.SubstitutionVisitor.go(SubstitutionVisitor.java:466) 
{code}

h2. Issue 2: result mismatch
{code}
  @Test public void testMatResultsMismatch() {
    CalciteAssert.that()
        .withMaterializations(
            HR_FKUK_MODEL,
            "m0",
            "select avg(\"salary\"), count(*), max(\"salary\"), 
sum(\"salary\"), min(\"salary\"), \"empid\" from \"emps\" group by \"empid\"",
            "m1",
            "select avg(\"salary\"), count(*), max(\"salary\"), 
sum(\"salary\"), min(\"salary\"), \"deptno\", \"empid\" from \"emps\" group by 
\"empid\", \"deptno\""
        )
        .query(
            "select count(*), \"empid\" from \"emps\" group by \"empid\" order 
by count(*) desc limit 50"
        )
        .enableMaterializations(true)
        .explainContains("EnumerableTableScan(table=[[hr, m0]])")
        .sameResultWithMaterializationsDisabled();
  }
{code}
Error is the same as in the previous case, but if using a workaround to disable 
MaterializedViewSubstitutionVisitor, there will be a result mismatch:
{code}
java.lang.AssertionError: 
Expected: "EXPR$0=1; empid=100\nEXPR$0=1; empid=150\nEXPR$0=1; 
empid=200\nEXPR$0=1; empid=110\n"
     but: was "EXPR$0=1; empid=100\nEXPR$0=1; empid=110\nEXPR$0=1; 
empid=150\nEXPR$0=1; empid=200\n"
<Click to see difference>
{code}

> Error while applying rule MaterializedViewAggregateRule(Project-Aggregate): 
> ArrayIndexOutOfBoundsException
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: CALCITE-3052
>                 URL: https://issues.apache.org/jira/browse/CALCITE-3052
>             Project: Calcite
>          Issue Type: Bug
>    Affects Versions: 1.19.0
>            Reporter: Anton Haidai
>            Priority: Major
>
> *Materialized views enabled:*
> # {{select avg(grade), count\(*), max(grade), sum(grade), min(grade), team 
> from students group by team}}
> # {{select avg(grade), count\(*), max(grade), sum(grade), min(grade), team, 
> faculty from students group by faculty, team}},
> *Query:*
> # {{select count\(*), team from students group by team}}
> *Error* (stacktrace is obtained using the current *master* branch: 
> "247c7d4f76"):
> {noformat}
> Caused by: java.lang.ArrayIndexOutOfBoundsException: 2
>       at 
> com.google.common.collect.RegularImmutableList.get(RegularImmutableList.java:60)
>       at org.apache.calcite.rex.RexBuilder.makeInputRef(RexBuilder.java:841)
>       at 
> org.apache.calcite.rel.rules.AbstractMaterializedViewRule$MaterializedViewAggregateRule.rewriteView(AbstractMaterializedViewRule.java:1507)
>       at 
> org.apache.calcite.rel.rules.AbstractMaterializedViewRule.perform(AbstractMaterializedViewRule.java:522)
>       at 
> org.apache.calcite.rel.rules.AbstractMaterializedViewRule$MaterializedViewProjectAggregateRule.onMatch(AbstractMaterializedViewRule.java:1776)
>       at 
> org.apache.calcite.plan.volcano.VolcanoRuleCall.onMatch(VolcanoRuleCall.java:208)
>       ... 71 common frames omitted
> {noformat}
> Reproducible only if both Materialization views listed are enabled: any 
> single one of these two could be successfully used with the query without any 
> errors. Looks like is is reproducible when AbstractMaterializedViewRule is 
> trying to rewrite one materialized view using the another materialized view.
> Currently, I'm trying to reproduce the issue in "MaterializationTest": 
> without a success so far, I'll update the ticket if I'll find a working way 
> to reproduce the issue in the test will be discovered.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to