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

    https://github.com/apache/incubator-quarks/pull/167#discussion_r71019197
  
    --- Diff: api/topology/src/main/java/quarks/topology/TWindow.java ---
    @@ -105,8 +129,85 @@ Licensed to the Apache Software Foundation (ASF) under 
one
          * @return A stream that contains the latest aggregations of 
partitions in this window.
          */
         <U> TStream<U> batch(BiFunction<List<T>, K, U> batcher);
    +
    +    /**
    +     * Declares a stream that is a continuous, sliding, 
    +     * timer triggered aggregation of
    +     * partitions in this window.
    +     * <P>
    +     * Periodically trigger an invocation of
    +     * {@code aggregator.apply(tuples, key)}, where {@code tuples} is
    +     * a {@code List<T>} containing all the tuples in the partition in
    +     * insertion order from oldest to newest.  The list is stable
    +     * during the aggregator invocation.  
    +     * The list will be empty if the partition is empty.
    +     * </P>
    +     * <P> 
    +     * A non-null {@code aggregator} result is added to the returned 
stream.
    +     * </P>
    +     * <P>
    +     * Thus the returned stream will contain a sequence of tuples where the
    +     * most recent tuple represents the most up to date aggregation of a
    +     * partition.
    +     *
    +     * @param <U> Tuple type
    +     * @param period how often to invoke the aggregator
    +     * @param unit TimeUnit for {@code period}
    +     * @param aggregator
    +     *            Logic to aggregation a partition.
    +     * @return A stream that contains the latest aggregations of 
partitions in this window.
    +     * 
    +     * @see #aggregate(BiFunction)
    +     */
    +    <U> TStream<U> timedAggregate(long period, TimeUnit unit, 
BiFunction<List<T>, K, U> aggregator);
    --- End diff --
    
    One of the issues I had with this in a previous system was that with many 
partitions the behaviour was not desired in that if the partition did not 
change the window still fired, thus wasting cpu cycles to produce the same 
result. Thus I wonder if it should be more along the lines of:
    
    Aggregation of window partition on any partition change with a minimum 
period of `period` between aggregations. 


---
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 [email protected] or file a JIRA ticket
with INFRA.
---

Reply via email to