[ 
https://issues.apache.org/jira/browse/FLINK-4856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15875843#comment-15875843
 ] 

ASF GitHub Bot commented on FLINK-4856:
---------------------------------------

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

    https://github.com/apache/flink/pull/3336#discussion_r102187482
  
    --- Diff: 
flink-contrib/flink-statebackend-rocksdb/src/main/java/org/apache/flink/contrib/streaming/state/RocksDBMapState.java
 ---
    @@ -382,11 +342,26 @@ private UV deserializeUserValue(byte[] rawValueBytes) 
{
                        this.rawValueBytes = rawValueBytes;
                        this.deleted = false;
                }
    -           
    +
    +           public void remove() {
    +                   deleted = true;
    +                   rawValueBytes = null;
    +
    +                   try {
    +                           db.remove(columnFamily, writeOptions, 
rawKeyBytes);
    +                   } catch (RocksDBException e) {
    +                           throw new RuntimeException("Error while 
removing data from RocksDB.", e);
    --- End diff --
    
    The intention of my comment about `RuntimeException` was not about changing 
the method signatures for throwing RocksDBExcepion. My suggestion was to only 
use a proper subclass of `RuntimeException`. We should avoid using 
`RuntimeException` directly, similar to how we should avoid throwing the class 
`Exception` directly. I know that there is some code in Flink that does not 
follow this, but I think it is better code style to stick with more appropriate 
subclasses.


> Add MapState for keyed streams
> ------------------------------
>
>                 Key: FLINK-4856
>                 URL: https://issues.apache.org/jira/browse/FLINK-4856
>             Project: Flink
>          Issue Type: New Feature
>          Components: DataStream API, State Backends, Checkpointing
>            Reporter: Xiaogang Shi
>            Assignee: Xiaogang Shi
>
> Many states in keyed streams are organized as key-value pairs. Currently, 
> these states are implemented by storing the entire map into a ValueState or a 
> ListState. The implementation however is very costly because all entries have 
> to be serialized/deserialized when updating a single entry. To improve the 
> efficiency of these states, MapStates are urgently needed. 



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to