[ 
https://issues.apache.org/jira/browse/FLINK-16322?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Flink Jira Bot updated FLINK-16322:
-----------------------------------
      Labels: auto-deprioritized-major auto-deprioritized-minor  (was: 
auto-deprioritized-major stale-minor)
    Priority: Not a Priority  (was: Minor)

This issue was labeled "stale-minor" 7 days ago and has not received any 
updates so it is being deprioritized. If this ticket is actually Minor, please 
raise the priority and ask a committer to assign you the issue or revive the 
public discussion.


> get wrong result after filter push down is applied in parquet table source
> --------------------------------------------------------------------------
>
>                 Key: FLINK-16322
>                 URL: https://issues.apache.org/jira/browse/FLINK-16322
>             Project: Flink
>          Issue Type: Bug
>          Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile)
>            Reporter: godfrey he
>            Priority: Not a Priority
>              Labels: auto-deprioritized-major, auto-deprioritized-minor
>         Attachments: parquet-1-1.parquet
>
>
> I get the wrong result when run the following query:
> source schema:
> first VARCHAR
> id INT
> score DOUBLE 
> last VARCHAR
> data: (parquet file is in the attachment)
> ("Mike", 1, 12.3d, "Smith"),
> ("Bob", 2, 45.6d, "Taylor"),
> ("Sam", 3, 7.89d, "Miller"),
> ("Peter", 4, 0.12d, "Smith"),
> ("Liz", 5, 34.5d, "Williams"),
> ("Sally", 6, 6.78d, "Miller"),
> ("Alice", 7, 90.1d, "Smith"),
> ("Kelly", 8, 2.34d, "Williams")
> query:
> SELECT id, `first`, `last`, score FROM ParquetTable WHERE score < 3
> the expected result size is 2, however the actual result size is 0. 
> If I disable filter push down, the result is correct. 
> I think there are two bugs: 
> the one is in ParquetTableSource#applyPredicate. AFAK, ParquetReader will 
> filter the row groups based on statistics not filter the record. So the 
> predicates passed to the applyPredicate method should not be removed.
> Even if I fix the first bug, the result is also wrong. So I guess there may 
> be a bug in other place.



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

Reply via email to