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

    https://github.com/apache/flink/pull/1255#discussion_r43953382
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/operators/shipping/OutputEmitter.java
 ---
    @@ -67,27 +67,11 @@ public OutputEmitter(ShipStrategyType strategy) {
         * @param comparator The comparator used to hash / compare the records.
         */
        public OutputEmitter(ShipStrategyType strategy, TypeComparator<T> 
comparator) {
    -           this(strategy, comparator, null, null);
    +           this(strategy, comparator, null);
        }
        
    -   /**
    -    * Creates a new channel selector that uses the given strategy 
(broadcasting, partitioning, ...)
    -    * and uses the supplied comparator to hash / compare records for 
partitioning them deterministically.
    -    * 
    -    * @param strategy The distribution strategy to be used.
    -    * @param comparator The comparator used to hash / compare the records.
    -    * @param distr The distribution pattern used in the case of a range 
partitioning.
    -    */
    -   public OutputEmitter(ShipStrategyType strategy, TypeComparator<T> 
comparator, DataDistribution distr) {
    -           this(strategy, comparator, null, distr);
    -   }
    -   
    -   public OutputEmitter(ShipStrategyType strategy, TypeComparator<T> 
comparator, Partitioner<?> partitioner) {
    -           this(strategy, comparator, partitioner, null);
    -   }
    -           
        @SuppressWarnings("unchecked")
    -   public OutputEmitter(ShipStrategyType strategy, TypeComparator<T> 
comparator, Partitioner<?> partitioner, DataDistribution distr) {
    +   public OutputEmitter(ShipStrategyType strategy, TypeComparator<T> 
comparator, Partitioner<?> partitioner) {
    --- End diff --
    
    I would keep the `DataDistribution` in the `OutputEmitter`. 
    We can later adapt this such that we can set the `DataDistribution` at 
execution time from the broadcasted set and avoid the partition ID assignment.
    
    For now, we can simply set an Integer distribution with boundaries 
1,2,3...,n. 


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

Reply via email to