[jira] [Updated] (FLINK-29489) Display issue when querying complex, deeply nested fields

["Martijn Visser (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/FLINK-29489?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Martijn Visser updated FLINK-29489:
-----------------------------------
    Component/s: Table SQL / API
                 Table SQL / Planner

> Display issue when querying complex, deeply nested fields
> ---------------------------------------------------------
>
>                 Key: FLINK-29489
>                 URL: https://issues.apache.org/jira/browse/FLINK-29489
>             Project: Flink
>          Issue Type: Bug
>          Components: Table SQL / API, Table SQL / Planner
>            Reporter: Justin
>            Priority: Major
>         Attachments: FlinkQueries.txt, flink-screenshot1.jpg, flink-screenshot2.jpg
>
>
> Using Flink 1.15, I’m observing some strange behavior when querying fields that have complex, deeply nested fields.
> The attached file, FlinkQueries.txt, has two create table statements.
> In the first table, I create a Flink table on top of the `tfmsStatusOutput` field. I have no issues querying this field using the SQL client or Java API.
> The second table, I create a Flink table on top of the `fltdOutput` field. This field is more complex and has deeply nested fields. Using the SQL client, when I run a simple bounded query such as `SELECT * FROM TBL2 LIMIT 1;` I get a stack trace dump in my display window and when I press any key on my keyboard it returns to the result window as if it's waiting for the results. Those are screenshots  (flink-screenshot1 and flink-screenshot2).
>  
> Using the Java API, I experience something similar. 
> It looks like it is going to return something good:
> Starting the SELECT...
> Job has been submitted with JobID 0b38a2c51e8357e5fef471b38ac839d0
> +----+--------------------------------+--------------------------------+--------------------------------+
> | op |                fltdOutput |
> but then it starts spitting out some form of diagnostics that doesn’t make much sense:
> /* 1 */public final class GeneratedCastExecutor$45 implements org.apache.flink.table.data.utils.CastExecutor {
> /* 2 */java.lang.StringBuilder builder$46 = new java.lang.StringBuilder();
> /* 3 */java.lang.StringBuilder builder$50 = new java.lang.StringBuilder();
> /* 4 */java.lang.StringBuilder builder$55 = new java.lang.StringBuilder();
> …
> /* 845 */java.lang.StringBuilder builder$7642 = new java.lang.StringBuilder();
> /* 846 */java.lang.StringBuilder builder$7654 = new java.lang.StringBuilder();
> /* 847 */java.lang.StringBuilder builder$7664 = new java.lang.StringBuilder();
> /* 848 */java.lang.StringBuilder builder$7672 = new java.lang.StringBuilder();
> /* 849 */public GeneratedCastExecutor$45() {
> /* 850 */}
> /* 851 */@Override public Object cast(Object _myInputObj) throws org.apache.flink.table.api.TableException {
> /* 852 */org.apache.flink.table.data.RowData _myInput = ((org.apache.flink.table.data.RowData)(_myInputObj));
> /* 853 */boolean _myInputIsNull = _myInputObj == null;
> /* 854 */boolean isNull$0;
> /* 855 */org.apache.flink.table.data.binary.BinaryStringData result$1;
> /* 856 */boolean isNull$2;
> /* 857 */org.apache.flink.table.data.binary.BinaryStringData result$3;
> /* 858 */boolean isNull$4;
> etc…
>  
> I couldn't find anything in the task manager logs.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)