[ 
https://issues.apache.org/jira/browse/FLINK-2203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14584637#comment-14584637
 ] 

ASF GitHub Bot commented on FLINK-2203:
---------------------------------------

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

    https://github.com/apache/flink/pull/831#discussion_r32371124
  
    --- Diff: 
flink-staging/flink-table/src/main/scala/org/apache/flink/api/table/typeinfo/RowSerializer.scala
 ---
    @@ -102,20 +119,39 @@ class RowSerializer(fieldSerializers: 
Array[TypeSerializer[Any]])
         val len = fieldSerializers.length
     
         val result = new Row(len)
    -    var i = 0
    -    while (i < len) {
    -      result.setField(i, fieldSerializers(i).deserialize(source))
    -      i += 1
    +
    +    var index = 0
    +    while (index < len) {
    +      val isNull: Boolean = source.readBoolean()
    +      if (isNull) {
    +        result.setField(index, null)
    +      } else {
    +        val serializer: TypeSerializer[Any] = fieldSerializers(index)
    +        result.setField(index, serializer.deserialize(source))
    +      }
    +      index += 1
         }
         result
       }
     
    +  private final val booleanSerializer = new BooleanSerializer()
    +
       override def copy(source: DataInputView, target: DataOutputView): Unit = 
{
         val len = fieldSerializers.length
         var i = 0
         while (i < len) {
    +      booleanSerializer.copy(source, target)
    --- End diff --
    
    I think it would be easier to do
    ```
    target.writeBoolean(source.readBoolean())
    ```
    here, instead of going through the extra abstraction of the 
BooleanSerializer.


> Add Support for Null-Values in RowSerializer
> --------------------------------------------
>
>                 Key: FLINK-2203
>                 URL: https://issues.apache.org/jira/browse/FLINK-2203
>             Project: Flink
>          Issue Type: Improvement
>          Components: Table API
>            Reporter: Aljoscha Krettek
>            Assignee: Shiti Saxena
>            Priority: Minor
>              Labels: Starter
>
> This would be a start towards proper handling of null values. We would still 
> need to add support for null values in aggregations.



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

Reply via email to