[ 
https://issues.apache.org/jira/browse/DRILL-4642?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15271455#comment-15271455
 ] 

ASF GitHub Bot commented on DRILL-4642:
---------------------------------------

Github user hsuanyi commented on a diff in the pull request:

    https://github.com/apache/drill/pull/489#discussion_r62117505
  
    --- Diff: 
exec/java-exec/src/test/java/org/apache/drill/TestFunctionsWithTypeExpoQueries.java
 ---
    @@ -213,7 +221,7 @@ public void tesIsNull() throws Exception {
         List<Pair<SchemaPath, TypeProtos.MajorType>> expectedSchema = 
Lists.newArrayList();
         TypeProtos.MajorType majorType = TypeProtos.MajorType.newBuilder()
             .setMinorType(TypeProtos.MinorType.BIT)
    -        .setMode(TypeProtos.DataMode.REQUIRED)
    +        .setMode(TypeProtos.DataMode.OPTIONAL)
    --- End diff --
    
    In fact, I meant to let the tests in this test-suite run with Limit-0 
optimization. However, that option is off by default. So I added 
    
https://github.com/apache/drill/pull/489/files#diff-0e0d4b8f1e54d2a7e8b6bc04d8606bf7R35
    
    However, that exposed an issue here:
    https://issues.apache.org/jira/browse/DRILL-4656
    
    So I think we should firstly turn this test off until we fix DRILL-4656


> Let RexBuilder.ensureType() mechanism take place during Rex conversion.
> -----------------------------------------------------------------------
>
>                 Key: DRILL-4642
>                 URL: https://issues.apache.org/jira/browse/DRILL-4642
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Query Planning & Optimization
>            Reporter: Sean Hsuan-Yi Chu
>            Assignee: Sean Hsuan-Yi Chu
>             Fix For: 1.7.0
>
>
> In DRILL-4372, the logic of ensuring same type is removed since, in some case 
> such as below, undesirable cast function will be added and cause failure.
> {code}
> SELECT * 
> FROM T 
> WHERE (cast(col1 as timestamp)  - to_timestamp(col2,'YYYY-MM-dd HH:mm:ss') < 
> interval 'X XX:XX:XX' day to second)
> {code}
> The fundamental reason for this behavior roots in Drill-Calcite [1], where 
> SqlNode WHERE is expanded to a new object but is not passed into validation 
> step.
> [1] 
> https://github.com/mapr/incubator-calcite/blob/DrillCalcite1.4.0-mapr-1.4.0/core/src/main/java/org/apache/calcite/sql/validate/SqlValidatorImpl.java#L3362



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to